Association of Ideal Cardiovascular Health in Youth with Cancer Risk in Adulthood: Cardiovascular Risk in Young Finns Study

Abstract

Background

Lifestyle factors may affect cancer risk. This study aimed to identify whether American Heart Association (AHA) Ideal Cardiovascular Health (ICH) score and its individual variables in youth associate with subsequent cancer incidence.

Methods

Study comprised of participants of the Cardiovascular Risk in Young Finns Study free of cancer at analysis baseline in 1986 (n=1873). Baseline age was 12-24 years and the follow-up occurred between 1986-2018.

Results

Among 1873 participants (mean age 17.3±4.1 years; 53.4% females at baseline), 72 incident cancer cases occurred during the follow-up (mean follow-up time 31.4±3.4 years). Baseline ICH score was not associated with future cancer risk (HR 0.96, 95% CI 0.78-1.12 per 1-point increment). Of individual ICH score variables, ideal physical activity (PA) was inversely associated with cancer incidence (age- and sex-adjusted HR 0.45 (0.23-0.88) per 1-category change [nonideal/ideal]), and remained significant in multivariable-adjusted model including also BMI, smoking, diet and socioeconomic status. A continuous physical activity index at ages 9-24 years and moderate to vigorous physical activity in youth were also related to decreased cancer incidence (p<0.05). BMI, smoking, diet, total cholesterol, glucose and blood pressure were not related to cancer risk. Of the dietary components, meat consumption was associated with cancer incidence (p=0.023).

Conclusions

These findings indicate that higher PA levels in youth associate with a reduced subsequent cancer incidence whereas AHA’s ICH score in youth does not.

Impact

This finding supports the efforts in promoting healthy lifestyle and encourages in physical activity during childhood yielding in subsequent healthier life.

Abbreviations

AHA

American Heart Association

AUC

area under the curve

BMI

body mass index

BP

blood pressure

CVD

cardiovascular disease

DBP

diastolic BP

ICH

ideal cardiovascular health

ideal PA

ideal physical activity

MVPA

moderate-to-vigorous-intensity physical activity

PA

physical activity

PAI

physical activity index

PAI AUC

cumulative PA area under the curve

SBP

systolic BP

Introduction

In 2010, the American Heart Association (AHA) put forth the concept of ideal cardiovascular health (ICH) as part of the 2020 Strategic Impact Goals for the organization (1). Simultaneously, it determined 7 metrics for monitoring ICH, consisting of 4 ideal health behaviors (nonsmoking, body mass index [BMI], physical activity [PA] and diet) and 3 ideal health factors (serum total cholesterol and glucose levels, and blood pressure [BP]). In this concept, recommendations for ideal health behaviors and factors for adults were given, and correspondent appropriate levels for children were provided. The original agenda was to improve cardiovascular health. However, thereafter the concept of ICH has shown to be relevant to several other diseases and overall mortality (1–6). Thus far, despite of the accumulating data on the association of ICH with lower disease burden and better health, studies have reported that less than 1% of adults to have ICH (2, 7). Moreover, the situation reported for US children younger than 12 years (8) or adolescents aged 12-19 years (9) has not been much more consoling, and ICH further decreases at later ages (9, 10).

Cancer is a major health burden with increasing incidence rates in both sexes (11, 12). Following cardiovascular diseases (CVDs), cancers are reported to be the number two cause of death worldwide (11, 13). Thus, efforts in exploring the primary cause and risk factors for increased cancer incidence, in addition to attempts to attenuate already known risk factors, have gained high interest. Today, the associations of lifestyle characteristics with cancer incidence risks are reported, including those included in AHA’s ideal cardiovascular health metrics (1, 14–16). The sum ICH score in general in addition to individual ICH components of ideal health behaviors (nonsmoking, adequate PA, normal BMI and healthy diet), has been associated with decreased cancer risk in some cohort studies of adult populations (17–20). The most relevant ICH associated individual cancers are lung, colorectal, and breast cancer. Based on prior studies, smoking and PA seem to play important individual roles in adults (17, 19–20). Nevertheless, whether ICH or some of its individual metrics already present in early life and adolescence associate with subsequent cancer risk has not been comprehensively investigated previously.

In this study, we sought to investigate whether ICH score and its individual metrics in adolescence/young adulthood associate with subsequent cancer incidence in the population-based Cardiovascular Risk in Young Finns Study cohort with over three decades of follow-up.

Materials and Methods

Participants

The Cardiovascular Risk in Young Finns Study (YFS) (21, 22) is an ongoing multicenter prospective longitudinal population-based follow-up study to investigate CVD risk factors in a Finnish cohort of individuals being randomly chosen from the national population register in five study areas. It has followed participants from childhood into adulthood nearly 40 years since the first baseline cross-sectional survey was conducted in 1980. Of the 4,320 randomly selected children and adolescents (aged 3 to 18 years in 1980; and born in 1962, 1965, 1968, 1971, 1974, and 1977; respectively) from the Finnish national population register, 3596 individuals participated across 5 Finnish University cities and their rural surroundings in 1980. Thereafter, since 1980, multiple follow-up studies have been conducted regularly, including 1986 (n=2799), 2001 (n=2283) (23), 2007 (n=2204) (24) and the latest of previous surveys, which was performed in 2011 (25) when 2063 participants (aged 34-49 years) from the original cohort participated. Each follow-up survey has collected clinical measurements, and in addition, questionnaires were used to collect lifestyle data. For the present study, the 1986 survey was chosen as the baseline, as it was the first year when follow-up of complete AHA’s ICH metrics, also including glucose measurements, was started. Of the possible participants, we excluded those lacking present cancer data and/or who have denied permission for their data usage for non-cardiovascular research (n=17). In addition, we excluded those participants lacking any of ICH metrics needed for lifestyle stratification (n=898).

In total, in the current study, 1873 participants in 1986 (age 12-24 years), n=1209 in 2001 (age 27-39), n=1118 in 2007 (age 33-45), and n=843 (age 37-49) in 2011 with complete data on ICH metrics in 1986 underwent analysis from 1986 through 2011. Data on cancer incidence was collected from 1986 until December 31^st, 2018 (mean follow-up time 31.4±3.4 years; range 0.9-32.9 years). During the follow–up, 57 (36.1±10.3 years; mean±SD; range 16.0-56.0 years) participants have died (cancer, n=8; other reasons, n=49). Among them, a questionnaire to assess PA was completed at baseline (1986) in childhood and adolescence (age range and participant number; 12-24 years; n=1818) and in adulthood in 2001, 2007, and 2011 (age range and participant number; 27-39 years, n=1417 in 2001; 33-45 years, n=1281 in 2007; and 37-49 years, n=1141 in 2011).

All participants and/or their legal guardians gave written informed consent, and the study was conducted in compliance with the Declaration of Helsinki and approved by the Joint Commission on Ethics of the University of Turku and the Turku University Central Hospital.

Cancer incidence

Cancer incidence data was identified from the Finnish national cancer registries, from the Care Register for Health Care and from national death records. ICD codes of primary and secondary cancers as well as the date of cancer diagnosis, cancer site of origin, and cancer histology were collected of each participant when available. Deaths and causes of deaths were collected from the Finnish national death records. Cancer death was defined as a participant diagnosed with cancer (previously confirmed or histologically confirmed by an autopsy), and in addition, when cancer was designated as a primary or secondary cause of death. Participants were classified into the following groups: no cancer or cancer; if cancer was diagnosed, participants were divided into further sub-groups according to cancer subtypes. After excluding participants with cancer diagnosis before 1986 (n=10), those lacking cancer information (n=17), those diagnosed with non-melanoma skin cancers only (n=16) and those lacking complete AHA ICH data (n=898) were excluded from the analysis, 72 incident cancer cases (61.1% female) occurred after 1986 among the studied cohort (n=1873 in 1986). Among them, the most common cancers were breast cancer (n=19, 100% female), melanoma (n=10), hematological (n=10), and gastrointestinal cancer (n=9). Supplementary Table S1 presents the characteristics of different cancer diagnoses until 2018. Supplementary Table S2 presents the age at cancer in chronological order among those 72 incident cancer cases presented during follow-up. For primary data analysis, all cancer cases were combined. In the occasion of multiple cancers, the earliest date of the primary cancer was used in the combined analysis for cancer incidence.

Ideal cardiovascular health

ICH was determined based on AHA-defined ideal cardiovascular health metrics including ideal health behaviors (BMI, nonsmoking, ideal diet based on recommendations, and PA at goal level) as well as ideal health factors (BP, total cholesterol level, and fasting plasma glucose), and for children, previously reported age- and sex-specific percentile cut-points for BP and BMI were used throughout the study (1, 3). Self-reported questionnaires were used to collect data on smoking, dietary habits, PA and parental history of disease burden. ICH metrics were obtained from all participants at baseline in childhood/adolescence (1986) and at subsequent visits in adulthood (2001, 2007, and 2011). A corresponding ICH score was generated from these individual health factors as previously reported (3, 4). For the ICH score, a value of 1 was given for each metric if the criterion for ideal was met; if a health factor or behavior did not meet this criterion, a value of 0 was assigned. This introduced an ICH score ranging from 0 to 7; a higher score indicates better ideal cardiovascular health. In this study, the ICH score at baseline refers to the number of ICH factors and behaviors present in 1986. Respectively, the ICH score at follow-up corresponds to the number of ideal health factors and behaviors present in 2011, unless otherwise mentioned. In the analyses, the ICH score was standardized for age and sex to ensure that the data were comparable between early life (childhood/adolescence criteria) and adulthood (adult criteria). To evaluate the differences between participants with or without complete AHA ICH data in 1986, we compared those with complete AHA ICH data (n=1873) in 1986, against those presented in original study cohort (1980) and without complete AHA ICH data in 1986 (n=1102). In addition, to study how the ICH score and its individual components changed overtime from 1986 to 2011, we selected those participants having all components of ICH both at 1986 and at 2011 surveys into analysis, resulting in total of 843 participants for comparison.

Health behaviors

Body mass index

Height and weight were measured, and BMI was calculated as weight in kilograms divided by height in meters squared. BMI was expressed as age- and sex-specific percentiles for participants <20 years of age and classified as ideal in childhood if BMI was <85^th percentile. In adults, the ideal BMI criterion was <25 kg/m² (1).

Diet

In youth, dietary data was obtained using a non-quantitative food frequency questionnaire (FFQ) to evaluate the consumption frequency of several food items (e.g., vegetables, fruit, meat, and fish) (26). When participants were <12 years old, dietary data was obtained from parents, and when participants were aged 12 years or older, the participants completed the FFQ by themselves with the aid of parents if needed (26, 27). In 1986 and 2001, an ideal diet was defined as having 2-3 of 3 ideal diet components (fruit and vegetables, fish and fish products, and soft drinks), and in 2007 and 2011, the ideal diet comprised 4-5 of the 5 dietary goals (consumption of fruit and vegetables, fish, whole grains, sugar-sweetened beverages, and the intake of sodium), where whole grains and sodium consumption were now added as a novel components of ideal diet to define new ideal diet, as previously described (1, 3). A more detailed FFQ available in 2007 and 2011 has been described previously (26, 27). In brief, to examine the frequency of meat, fish, vegetable, and fruit consumption, the participants were asked to fill in a questionnaire on habitual dietary choices, formulating six response categories (1=daily, 2=almost every day, 3=a couple of times per week, 4=about once a week, 5=a couple of times per month, 6=more seldom). For the following analysis, the response categories were converted into times of consumption per week (1, >9.5; 2, >6.3; 3, >3; 4, >1.2; 5, >0.3; 6, >0.1; respectively).

Smoking

Information on smoking habits was obtained using questionnaires in connection with the study visits. When the participant was ≥12 years old, data was collected in a solitary room to ensure confidentiality. In children, smoking status was defined as ideal if she/he had never smoked a whole cigarette, and was otherwise non-ideal. In adults, smoking status was classified as ideal if the participant had never smoked or had quit smoking >12 months ago; otherwise smoking status was classified as non-ideal (current smoker) (1).

Physical activity

Self-reported questionnaires, completed in connection with a medical examination visit, were used to assess baseline PA, as reported previously (28). The acquired data comprised information on the duration, intensity, and frequency of PA. According to AHA’s definition (1), ideal PA at the goal level was achieved in adults when reaching ≥150 minutes per week of moderate intensity physical activity or 75 minutes per week of vigorous intensity aerobic PA or combination. For children, the criterion for ideal PA is at least 60 minutes of moderate to vigorous activity per each day of the week. When the criteria, respectively, was not met, PA was considered non-ideal. Thereafter, those participants reaching the ideal PA level were given 1 point (ideal PA; n=374), and those who did not reach the ideal PA level were given 0 points (non-ideal PA; n=1499).

Additionally, a physical activity index (PAI) was calculated. The PAI in adolescence and early adulthood (i.e., youth) was generated based on information obtained from participants aged 9-24 years in 1986, as described previously (28). The PAI was age-standardized to ensure that the data is comparable between the groups studied. The PAI comprised information on intensity and frequency of leisure-time PA, participation in organized PA in for example sport clubs, participation in competitive sports, school physical education grades and type of school commute, and typical leisure-time activity. The PAI in 1986 ranged from 5 to 14, with 5 indicating the lowest and 14 the highest PA level. The PAI in adulthood – i.e., in 2001, 2007 and 2011 – was similarly generated (29). In adulthood, PAI ranged between 5-15, with 5 indicating the lowest and 15 the highest PA level. The adulthood PAI included information on the frequency of PA, intensity of PA, average time spent in individual activity, frequency of vigorous PA, average duration of PA sessions, and participation in organized PA. PA questionnaire has been previously validated in this study cohort by measuring its correlation to objective measures (accelometers and pedometers) (30). In the questionnaires, data was collected on time of vigorous training per week with a scale from 1 to 6 (i.e., 1: none; 2: approx. 30 min; 3: approx. 60 min; 4: approx. 120-180 min; 5: approx. 240-360 min; and 6: > 360 min of vigorous physical training per week). From this data three new subgroups were formed: Group 1 represents those participants with 30 minutes or less vigorous physical activity per week (reference group); group 2 represents approx. 60-180 minutes; group 3 represents approx. 240 minutes or more vigorous physical activity per week.

To utilize all available repeatedly measured exposure data and to generate the cumulative physical activity index data indicating a long-term exposure of PA, in model was used repeated measurements of physical activity (PA) data in various time-points and used cumulative area under the curve (AUC) data of PA generated by calculating estimates of area under curve (av.=9, SD=2; PAI index range 5-14); subject specific curves for PAI were then estimated using mixed model regression splines as reported previously (31). Resulting cumulative PAI index (PAI AUC) better illustrates the average PA during childhood. That further resulted in cumulative PAI AUC index describing PA between ages 9 to 24 years (youth). Additionally, to study whether limiting the observation time for PA cumulation in childhood and adolescence up to 18 years of age (9-18 years), instead of up to 24 years (9-24 years), would affect the result, we analyzed the association of cumulative PAI AUC data during 9-18 years of age with cancer incidence.

Ideal health factors (blood pressure, serum total cholesterol and glucose)

Blood pressure

Right arm BP was measured at rest according to guidelines both in childhood and adulthood from the brachial artery using a random-zero sphygmomanometer in 1986 and thereafter (27). The average of 3 readings was used in analysis. The BP status in childhood was classified as ideal when systolic blood pressure (SBP) was <90^th percentile and diastolic blood pressure (DBP) was <90^th percentile or non-ideal when SBP was ≥90^th percentile or DBP was ≥90^th percentile. Correspondingly, in adulthood, BP was considered ideal when SBP was <120 mmHg and DBP was <80 mmHg or as non-ideal when SBP was ≥120 mmHg or DBP ≥80 mmHg (1).

Venous sample collection and serum variables

Venous sample collection and measurement of serum cholesterol, triglyceride and glucose concentrations were performed as previously described (3, 27). In brief, venous blood samples were drawn after overnight fasting. Concentrations of serum total cholesterol, HDL-cholesterol, and triglycerides were analyzed using standard enzymatic methods, and LDL-cholesterol was analyzed using Friedewald’s equation (32). Serum total cholesterol was considered ideal when having untreated total cholesterol <4.40 mmol/L (<170 mg/dL) in childhood and <5.17 mmol/l (<200 mg/dL) in adulthood, and non-ideal if ≥4.40 mmol/L (≥170 mg/dL) in childhood and ≥5.17 mmol/l (≥200 mg/dL) in adulthood (1). Similarly, glucose concentrations were analyzed enzymatically with a clinical chemistry analyzer (AU400; Olympus) and considered as ideal in children and adults when the concentration was <5.6 mmol/L (<100mg/dL) and non-ideal if ≥5.6 mmol/L (≥100mg/dL) (1). Glucose concentrations were measured in 1986, 2001, 2007 (22), and 2011 (25).

Socioeconomic factors

In childhood, the length of parent’s education was considered an indicator of socioeconomic study year status (33), and data on parental study years were assessed by a self-report questionnaire. Similarly, socioeconomical income status was based on parents combined 1985 income tax registry information, and then families were divided into 7 categories based on their total income/year.

Statistical analysis

Distribution of each variable was examined using histograms and normal probability plots. In some variables, visual inspection of the regression graph curve demonstrated an approximately linear association of variables with cancer incidence, although the relationship of indices with incident cancer statistically contravened the assumption of linearity. Due to skewed distribution, BMI was log-transformed for statistical analysis. Continuous variables are presented as mean±SD and as median and interquartile ranges (IQR) in cases of non-normality, and categorical data/binary variables are presented as frequencies and percentages. Sex- and age-adjusted risk variables and PAI were entered into linear or logistic regression analyses to evaluate the independent association of lifestyle risk factors and traditional risk factors with cancer incidence. Then, to study how ICH score and its individual components or various PA measures were associated with combined cancer incidence until 2018, we examined hazard ratios and p-values using Cox proportional hazard models. The analysis was adjusted for sex and age, and additionally for BMI, smoking, dietary components, and socioeconomical factors in multivariable-adjusted models, when mentioned. Pearson correlation analysis, adjusted for sex and age, were performed for comparison of 1986, 2001, 2007 and 2011 physical activity indexes against each other. Statistical tests were performed using SAS version 9.4 (SAS Institute, Inc., Cary, NC). Statistical significance was inferred at a 2-tailed p <0.05, 95% confidence intervals (CIs) and HR are presented where relevant.

Results

Youth characteristics among those with and without cancer in adulthood

Altogether 1873 participants (mean age at baseline in 1986: 17.3 ± 4.1 years; 53.4% females) were followed up for 31.4±3.4 years (mean±SD; range 0.9-32.9 years) between 1986-2018. Among them were 72 incident cancer cases (43.9±7.9 years; mean±SD; range 21.4-56.9 years; Supplementary Table S2). Table 1 presents baseline characteristics of the participants with or without incident cancer during follow-up until adulthood. Those with cancer later in life were older and had a lower PAI than those without cancer, whereas no differences in sex, BMI, total, HDL- or LDL-cholesterol, triglycerides, and glucose, systolic and diastolic BP or smoking habits were found at baseline. Furthermore, to analyze the association of the consumption of distinct dietary components in more detail, we studied the association of eating fruits, vegetables, fish, and meat with cancer. Among these food items, meat consumption was associated with cancer.

Table 1. Youth Characteristics in 1986 (Means [SDs] or Percentages [%]) and Proportions of Components of Cancer Incidence According to Adult Cancer Status.

	Cancer Status at Follow-up
	No	Yes	P value
N (1876)	1801	72
Sex (N/%)
Males/Females	845 (46.9)/956 (53.1)	28 (38.8)/44 (61.2)	0.23
Age (years)	17.2±4.1	19.4±4.0	<0.0001
BMI (kg/m2)	20.8±3.2	21.4±3.2	0.73
Total cholesterol (mmol/l)	4.9±1.0	4.9±0.9	0.78
LDL-cholesterol (mmol/l)	3.0±0.9	3.0±0.8	0.82
HDL-cholesterol (mmol/l)	1.5±0.3	1.5±0.3	0.51
Triglycerides (mmol/l)	1.0± 0.4	1.0±0.5	0.41
Glucose (mmol/l)	4.7±0.7	4.6±0.5	0.56
Systolic BP (mmHg)	116.6±12.3	117.3±13.4	0.29
Diastolic BP (mmHg)	66.6±10.1	68.5±9.7	0.96
Physical activity index (5-14)	8.8±2.1	8.0±1.8	0.029
Dietary variables
consumption (times/week)
Fruit	7.0 (2.7)	6.7 (3.3)	0.72
Vegetable	6.4 (2.8)	5.9 (2.9)	0.56
Fish	1.1 (1.2)	0.9 (0.9)	0.28
Meat	5.0 (2.4)	5.5 (2.8)	0.023
ICH Score (0-7)	3.7 (1.1)	3.8 (1.4)	0.71
Youth smoking* (N/%)
Never/Yes	1312 (72.9)/488 (27.1)	48 (66.7)/24 (33.3)	0.48

P values are from age- and sex-adjusted ANCOVA test or age- and sex-adjusted logistic regression model. Abbreviations; BMI, body mass index; LDL, low density lipoprotein; HDL, high density lipoprotein; BP, blood pressure; ICH, ideal cardiovascular health. *If one has smoked ever daily before age 24 years.

Youth ideal cardiovascular metrics and other risk factors in predicting cancer in adulthood

The age- and sex-adjusted HRs for incident cancer across youth ICH score and its individual components are presented in Table 2. The ICH score at baseline (1986) in the youth (HR 0.96, 95% CI 0.78-1.12 per 1-point increment in score) and of its individual components BMI, serum total cholesterol and glucose, blood pressure, nonsmoking and diet were not associated with cancer incidence whereas AHA ideal physical activity was inversely associated with later cancer incidence (HR 0.45, 95% CI 0.23-0.88 per 1-category change [nonideal, ideal]). Results were unaltered when including all the components together in a model showing that AHA ICH PA component associates with cancer (HR 0.44, CI 0.22-0.88, [p=0.02] whereas the rest ICH components do not (p>0.05). Next, we performed analyses to study whether AHA ICH score associates with cancer incidence also on later time-points (i.e., 2001, 2007, and 2011). There were no statistically significant association between ICH (age- and sex-adjusted) and cancer at these time-points. In 2001 HR was 0.87 (95% CI 0.73-1.03, p= 0.10) for a 1-point increment in AHA ICH score; in 2007 HR was 0.97 (0.83-1.14; p= 0.73); and in 2011 HR was 1.04 (0.87-1.24; p= 0.66).

Table 2. Age- and Sex-adjusted HRs and their 95% CIs for Youth Risk Factors (Year 1986) in Predicting Adult Cancer.

	HR (95% CI)	P value
AHA score	0.96 (0.78-1.12)	0.66
Ideal health metrics
Total cholesterol	0.99 (0.60-1.60)	0.95
Blood pressure	1.29 (0.75-2.21)	0.37
Glucose	0.73 (0.18-3.01)	0.66
Nonsmoking	0.79 (0.47-1.34)	0.38
Physical activity	0.45 (0.23-0.88)	0.019
Diet	1.40 (0.84-2.36)	0.20
BMI	1.02 (0.53-1.95)	0.95

HRs for 1-point increment in AHA score variable and one category change in categorical study variables. P values are from age- and sex-adjusted Cox’s regression analyses. N=1873 in all categories. AHA ICH was modelled in its own model and its each health metric components were each modeled separately in own models.

The association of AHA’s physical activity component with cancer incidence remained significant also after simultaneously adjusted with confounding factors age, sex, BMI, smoking and meat consumption (HR 0.46, 95% CI 0.24-0.92) and results were not altered, after simultaneously adjusted also with socioeconomic factors (HR 0.46, 95% CI 0.23-0.91) either for parental study years or income status. Sex- and age-adjusted socioeconomical status (either parental study years or income status) did not predict incident cancer in adulthood.

Table 3 presents results from comparison between those having complete ICH data in 1986 (n=1873) against those participants in 1980 survey not having complete ICH data in 1986 (n=1102). Those without complete data were older, more of them were males, they had higher triglycerides, glucose, diastolic blood pressure and they consumed less fruit. There were, however, no differences in PA index (PAI), meat consumption, smoking or cancer incidence. Then we analyzed, how participants’ ICH score changed from 1986 survey to 2011 survey. Table 4 presents data of ICH score and its components and their change from 1986 through 2011 for those who had complete ICH data on both surveys (n=843). Comparison of those participants having all ICH components at 1986 against 2011 evaluation of ICH components, demonstrates that in 2011 participants had less ideal BMI, diet, glucose, and blood pressure, and in contrast, more ideal PA, smoking, and total serum cholesterol when compared to 1986. Table 5 depicts the incidence of cancer among those with AHA’s ideal PA (2.9%; n=11/374) and those who did not reach the goal (i.e., non-ideal PA, 4.1%; n=61/1499), and indicates a higher cancer incidence among those with non-ideal PA compared with those with ideal PA (p=0.034) after adjusted with sex, age, smoking, BMI, meat consumption and socioeconomic factors (parental study years) in same model.

Table 3. Youth Characteristics in 1980 (Means [SDs] or Percentages [%]) According to AHA’s ICH Status in 1986.

	ICH status available at 1986
	No	Yes	P value
N (3546)	1102	1873
Sex (N/%)
Males/Females	577 (52.4)/525 (47.6)	873 (46.6)/1000 (53.4)	0.0017
Age (years)	12.8±3.9 (n=1102)	11.3±4.1 (n=1873)	<0.0001
BMI (kg/m2)	18.8±3.3 (n=1102)	17.9±3.0 (n=1863)	0.53
Total cholesterol (mmol/l)	5.3±1.0 (n=1095)	5.3±0.9 (n=1863)	0.69
LDL-cholesterol (mmol/l)	3.4±0.9 (n=1095)	3.4±0.8 (n=1863)	0.66
HDL-cholesterol (mmol/l)	1.6±0.3 (n=1095))	1.6±0.3 (n=1863)	0.27
Triglycerides (mmol/l)	0.71± 0.4 (n=1095)	0.65±0.3 (n=1863)	0.02
Glucose (mmol/l)	4.9±2.4 (n=72)	4.7±0.7 (n=1873)	0.013
Systolic BP (mmHg)	115.8±11.7 (n=1101)	113.0±11.2 (n=1867)	0.12
Diastolic BP (mmHg)	69.7±9.7 (n=1093)	68.3±9.5 (n=1864)	0.044
Physical activity index (5-14)	9.0±1.8 (n=916)	9.1±1.8 (n=1358)	0.12
Dietary variables consumption (times/week)
Fruit	6.3 (2.9) (n=1088)	7.0 (2.8) (n=1858)	0.0003
Vegetable	5.9 (3.0) (n=1091)	6.3 (2.9) (n=1856)	0.061
Fish	1.2 (1.2) (n=1086)	1.1 (1.2) (n=1857)	0.095
Meat	4.9 (2.5) (n=1092)	5.0 (2.5) (n=1861)	0.99
Youth smoking* (N/%)
Never/Yes	749 (74.8)/252 (25.2)	1380 (74.3)/477 (25.7)	0.39
Incident Cancer (N/%)
No/Yes	1061 (96.3)/41 (3.7)	1801 (96.2)/72 (3.8)	0.43

Table presents participants aged 6 years or older in 1980. P values are from age- and sex-adjusted linear or logistic regression analyses. BMI was log-transformed for statistical analysis. Glucose was from 1986.

Abbreviations; AHA, American Heart Association; BMI, body mass index; LDL, low density lipoprotein; HDL, high density lipoprotein; BP, blood pressure; ICH, ideal cardiovascular health in 1986; PAI, physical activity index.

^*If one has smoked ever daily between 12-18 years of age.

Table 4. Change in AHA ICH Score and ICH’s Individual Components (Means [SDs] or Percentages [%]) Overtime from 1986 through 2011.

	Year
	1986	2011
ICH Score (0-7; N)	3.9±1.1 (843)	3.6±1.5 (843)
ICH Score Components Nonideal/ideal (N/%)
Health Behaviors
BMI	100 (11.9)/743 (88.1)	459 (54.5)/384 (45.5)
Physical activity	652 (77.3)/191 (22.7)	344 (40.8)/499 (59.2)
Diet	628 (74.5)/215 (25.5)	792 (93.9)/51 (6.1)
Nonsmoking	510 (60.5)/333 (39.5)	154 (18.3)/689 (81.7)
Health Factors
Serum Total Cholesterol	482 (57.2)/361 (42.8)	418 (49.6)/425 (50.4)
Glucose	11 (1.3)/832 (98.7)	248 (29.4)/595 (70.6)
BP	268 (31.8)/575 (68.2)	433 (51.4)/410 (48.6)

AHA defined ideal cardiovascular health [ICH] score and its subcomponents consisting of 4 behaviors (nonsmoking, body mass index [BMI], physical activity and diet) and 3 ideal health factors (serum total cholesterol and glucose levels, and blood pressure [BP]). For ICH subcomponents in 1986, participant was assigned ideal (given 1 point) if: BMI, <85^th percentile; PA, physical activity >7 hours/week; Diet, having 2-3 of 3 ideal diet components (fruit and vegetables, fish and fish products, and soft drinks); Nonsmoking, who has never smoked; Cholesterol, total cholesterol<4.4 mmol/L (<170 mg/dL); Glucose, blood glucose <5.6 mmol/L (<100mg/dL) + w/o diabetes mellitus; BP, SBP<90^th percentile and DBP<90^th percentile; and otherwise assigned as unideal (given 0 point). For ICH subcomponents in 2011, participant was assigned ideal (1 point) if: BMI, <25 kg/m²; PA, physical activity >1 hours/week vigorous intensity, or 2-3 hours/week moderate intensity PA; having 4-5 of 5 ideal diet components (fruit and vegetables, fish and fish products, intake of salt, whole grains and sugar sweetened beverages); Nonsmoking, have never smoked or stopped smoking; Cholesterol, total cholesterol <5.17 mmol/l (<200 mg/dL) + no cholesterol lowering medication; Glucose, <5.6 mmol/L (<100mg/dL) + no glucose lowering medication; BP, SBP<120 mmHg and DBP <80 mmHg + no BP lowering medication; and otherwise assigned as unideal (0 point). ICH score in 1986 and 2011 was a sum score of all 7 components (range 0-7), where a higher score indicates better ideal cardiovascular health.

Table 5. Relationship between Different AHA’s Ideal Cardiovascular Health (ICH) Component Ideal Physical Activity (PA) Groups and Adult Cancer Incidence.

	Cancer Status
	No Cancer (N, %)	Cancer (N, %)	P value
AHA’s Physical activity	1801 (96.2)	72 (3.8)
Nonideal (N=1499)	1438 (95.9)	61 (4.1)	0.034*
Ideal (N=374)	363 (97.1)	11 (2.9)

Cancer incidence (N/%) in different ideal PA groups (ideal vs non-ideal). n/N= 61/1499 for cancer incidence in non-ideal and 11/374 in ideal PA groups.

^*P for HR 0.034 (PA predicting cancer; sex, age, smoking [if one has ever smoked between age of 12-18 years], BMI, meat consumption, and socioeconomic factor [parental study years] adjusted model).

Youth physical activity in predicting cancer in adulthood

Of the additional PA indices studied, PAI in 1986 predicted independently cancer incidence (age- and sex-adjusted analysis; HR 0.86, 95% CI 0.76-0.98 per 1-point increment in index; n=1818; p=0.025). The association of PAI with cancer incidence remained significant also after simultaneously adjusted with confounding factors age, sex, BMI, smoking and meat consumption (HR 0.88, 95% CI 0.77-0.998) and after adjusted also with socioeconomic factors (HR 0.87, 95% CI 0.77-0.998). Also, meat consumption remained significant in this multivariable model using BMI, smoking, diet and socioeconomic factor (parental study years; HR 1.14, 95% CI 1.04-1.25). In addition, moderate to vigorous physical activity was associated with decreased cancer incidence (Supplementary Table S3).

To confirm the association of adherence to AHA-recommended ideal PA levels in childhood with lower cancer incidence, we evaluated the association of an alternative definition of physical activity, namely the cumulative physical activity index (PAI AUC) data collected over several time-points and thus reflecting better the average PA levels during childhood between 9-24 years of age. Cumulative physical activity index between 9-24 years (i.e., youth) predicted independently cancer incidence (age- and sex-adjusted analysis; HR 0.71, 95% CI 0.54-0.93 per 1-SD increase in index; n= 1873; p=0.015). Further, combined adjustment with BMI, smoking and meat consumption in multivariable model analysis did not change the result (HR 0.71, 95% CI 0.54-0.95) neither did adjustment with socioeconomic factors (parental study year; HR 0.71, 95% CI 0.53-0.95). Results using other socioeconomical factor (income status) were similar (HR 0.71, 95% CI 0.54-0.95).

In the light of that the definition of age limits for childhood and youth might be somewhat ambiguous (34) and some studies may limit youth to participants aged <18 years of life, we limited some part of the analyses to study association of PA cumulation on cancer incidence by concentrating on subsample of participants being younger than 18 years of age during PA, independent of the participant’s age at incident cancer. In this model, cumulative physical activity index (PAI AUC) data between 9-18 years predicted independently cancer incidence (age- and sex-adjusted analysis; HR 0.71, 95% CI 0.53-0.94 per 1-SD increase in index; n= 1873; p=0.017). Further adjustment with BMI, smoking and meat consumption (HR 0.71, 95% CI 0.53-0.95) or socioeconomic factor (parental study years; HR 0.71, 95% CI 0.52-0.95) did not alter the result. In addition, when the analysis of association of PA with cancer was limited to consider only participants being aged >18 years at the time of incident cancer, we found that the both AHA defined ideal PA in youth (HR 0.41, 95% CI 0.21-0.79 per one category change; n=1869; p=0.0084) as well as PAI in 1986 (HR 0.86, 95% CI 0.76-0.98 per 1-point increment in index; n=1814; p=0.020) were inversely associated with the risk of cancer when age-and sex-adjusted and, in addition, results remained similar for AHA defined ideal PA in youth (HR 0.44, 95% CI 0.22-0.86; n=1833; p=0.016), as well as PAI in 1986 (HR 0.87, 95% CI 0.77-0.995; n=1778; p=0.042) when adjusted also with BMI, smoking, meat consumption and socioeconomical status (parental study years).

Correlation analysis showed that during adulthood sex- and age-adjusted PAI in 2001 correlated with PAI in 2011 (r= 0.44, n=986; p<0.0001), PAI in 2007 with PAI in 2011 (r= 0.58, n= 1009, p<0.0001), and PAI in 2001 with PAI in 2007 (r=0.52, n= 1115, p<0.0001). Respectively, PAI in childhood correlates relatively well with PAI in adulthood (PAI in 1986 vs PAI in 2001; r 0.33; n=1375; p<0.0001) demonstrating that PA is durable and stable among adulthood and by aging.

Discussion

Our findings from this population-based prospective study suggest that non-ideal physical activity in youth (i.e., childhood to early adulthood; 9-24 years of age) associates with subsequent cancer incidence. To the best of our knowledge, this is the first prospective study to report that non-adherence to AHA guideline recommendations for physical activity from childhood to early adulthood is associated with an increased risk of cancer in adulthood.

The cumulative risk for cancer is high, overall lifetime risk for 0-74 years being 20.2%, and cancer is estimated to become number one cause of death over CVD already in 2060 (13). Many factors, such as race, age, sex, genetics, BMI, family history of cancer, history of tobacco use, nutrition, diet, obesity, and lack of physical activity, are considered to play an important role in cancer incidence (16, 35). Many of these risk factors (like obesity, age, smoking, dietary patterns and inadequate physical activity) are shared between CVD and some cancers (14, 16, 20, 36). However, whether physical activity in early life plays a pivotal role in cancer risk in adulthood, has not been demonstrated.

It is noteworthy that in this study, we could not see an association of youth AHA indices in general with cancer, as previously reported by others among adults (17, 19–20). Our data, presenting the inverse association of achieved AHA defined ideal physical activity goal with cancer incidence, is, however, in line with some previous data on adults showing an inverse association between PA and incident cancer. Physical activity is known to associate with lower cancer incidence in adults (19, 36–39). For example, in adults, PA may reduce the risk of breast cancer (37). However, some studies contradict this, and PA in adolescence in 14-21 years old women was not significantly associated with later breast cancer risk, although some suggestive inverse association was observed (40). In addition, another large ARIC study cohort survey (5807 men and 7252 women; 45-64-years) evaluating the effect of physical activity on lifetime risk of cardiovascular disease (CVD) and cancer showed that achievement of PA level recommended by the World Health Organization (WHO) and comparable to that of AHA’s recommendations, was not associated with total cancer, although PA was associated with a lower lifetime risk of CVD (36). Possibly the noticed difference between this ARIC study (consisting of middle-aged participants at baseline) and that of ours (consisting of young children and adolescents at baseline) may reflect the difference in the demographic age structure between these two cohorts.

In this study, we found using multiple PA measures, that the adherence to AHA-recommended ideal PA levels in baseline was associated with lower cancer incidence. The association of physical activity with lower cancer incidence was confirmed by other variables illustrating physical activity than direct adherence to AHA-recommended ideal PA levels, namely by using various previously described physical activity indexes (28). Finally, the association of PA with lower cancer incidence was confirmed by using an alternative definition of physical activity, namely the cumulative physical activity index (PAI AUC) data (31) collected over the several time points, reflecting the PA levels between ages 9 to 24 years, or respectively, between 9-18 years of age, both confirming the inverse association of PA with cancer incidence and thus better illustrating average PA during childhood and/or adolescence. Importantly, as cancer progression is a complex multiform event, the association of PA with cancer incidence remained significant also when simultaneously adjusted with multiple confounding variables namely age, sex, BMI, smoking, diet and socioeconomical factors, thus supporting PA’s independent role in cancer progression. These approaches consolidated inverse association of increased physical activity with cancer incidence and the prediction of cancer with PA. Utilization of average cumulative physical activity is also beneficial because many self-reporting queries are known to be subject to reporting error bias (35) reflecting the difficulties in the objective measurement of physical activity. Moreover, when monitoring period of cancer incidence was limited to participants >18 years of age, the association of physical activity with decreased cancer incidence remained significant. Results might reflect the phenomenon that in the case of cancer, a certain minimum time from either a cancer-promoting trigger, or, respectively, from the preventative measure of action occurring at tender age, to the measured endpoint in adulthood is needed and may thus minimize the number of randomly occurring events, at least at the beginning of the follow-up. In addition, an analysis to study association of moderate to vigorous physical activity with cancer incidence demonstrated that physical activity, either present in childhood/adolescence or early adulthood, or both, negatively associates with incident cancer and leads to cancer incidence reduction compared to those without physical activity at any age. This importantly suggests that efforts put to increase PA at any age in life may yield an advantage in cancer prevention in addition to improving cardiovascular health. In fact, this is reciprocally supported by reports that sedentary lifestyle attributes are likewise known to increase cancer burden and CVD (15, 35).

In the present study, we noticed an association of meat consumption with cancer. This is in line with previous data, which have shown that meat consumption associates with both mortality and cancer, especially colorectal cancer (41) but in this study, that could not be analyzed in more details due to low number of certain cancer subtypes. Interestingly, we did not observe an association of BMI or tobacco use with cancer incidence in our rather young study cohort, yet both BMI and tobacco use are well-known risk factors for cancer (16, 35). Recently, childhood obesity was shown to correlate with overall adult cancer mortality in a large i3C consortium study involving 7 international longitudinal cohorts and 21,012 children from the USA, Australia, and Finland (42). However, also in that i3C consortium study, the Cardiovascular Risk in Young Finns Study cohort did not demonstrate an association of childhood BMI with adult cancer (42) probably reflecting some unrevealed differences between the various cohorts. Nevertheless, PA also indirectly contributes to BMI and weight, in addition to many other factors, and may thus further support reduction of cancer risk via reduction of excess weight gain, oxidative stress state, chronic low-grade inflammation, and overstimulation of endogenous sex hormones (43). It is also important to note that obesity has increased during the last decades (43) although it was not as common in 1986 as it is today. Regarding smoking, in the adult study showing an association of ICH score and its PA component with cancer (17), the trend of lower cancer incidence associated with a larger ICH metrics score was attenuated when smoking was removed from the sum of ideal health metrics. However, yet the prevalence of smoking at study baseline in our cohort was relatively high compared to that of today, the lack of association of smoking with cancer incidence in our study may partly reflect the fact that cancers represented in our cohort were heterogeneous in their types and represented with only 2 lung cancer cases. Second, our study participants were still relatively young, which results in rather low cancer incidence at the population level and is thus reflected in the relatively low number of cancer cases seen in our study cohort. It is likely that, in the coming decades, cancer incidence may rise due to aging of the study cohort, and may thus also reveal the long-term impact of smoking and obesity in this cohort. Notably, considering our cohort participants’ rather young age at cancer diagnosis (43.9±7.9, 21.4-56.9 years; mean±SD, range), we can fairly designate vast majority of them as ‘early-onset cancers’ i.e. definition specified for example for colorectal cancers diagnosed in patients under the age of 50 years (44) and breast cancer at age of 18-45 years (45).

The value of preventative measures for cardiovascular health and cancer is well known. It has been suggested that for example nearly 80% of CVD could be prevented through lifestyle risk factor modifications (14). Respectively, physically fit men have lower cancer mortality rates than unfit men even when presented with similar adiposity measures (46). Additively, maintaining ideal levels of cardiovascular health factors from childhood to adulthood, that is, by achieving a good lifestyle as a child, is considered superior to medical intervention (1). Thus, it is tempting to speculate that early intervention initiatives made to improve adherence to ideal cardiovascular health indices in childhood may partly contribute to attenuating cancer risk even at older ages. This data endorses that educational efforts once invested in growing youth, in order to promote physical activity, may be later harvested as enforced public health. In addition, as childhood PA is also known to predict PA in adulthood (28), the investments made in childhood behavioral education may be assumed to be durable into adulthood. In this study, among our cohort participants, evaluation of how the ICH score and its individual components changed overtime from 1986 through 2011 demonstrated that, although participants in 2011 had less ideal BMI, diet, glucose and BP, they yet had maintained or even enhanced their PA level, they smoked less and serum total cholesterol profile was more ideal, the last possible reflecting healthier dietary choices. In addition, the correlation analysis of participants’ PAI between various surveys within childhood and adulthood, showed that PA during life-span is stable within the adulthood, and in addition, that PA seems to be also durable from childhood into the adulthood in our cohort participants. For adults, it has been recommended that any PA is better than nothing for the reduction of cancer risk among people aged >65 years (35). In addition, a large population-based prospective adult cohort in the United States and Europe, consisting of 661,137 men and women with a median age of 62 years (range, 21-98 years), demonstrated beneficial dose-response effects of PA against mortality due to cardiovascular disease and cancer. Besides, there is no absolute upper risk limit for beneficial PA in adults (47). Importantly, as long as there may not exist evidence or agreement of the absolute minimum level of adequate PA in childhood or adolescence to yield positive health effects later in life, it is good to adhere to AHA’s guidelines of recommended minimum levels that are easily exploitable in daily life.

The major strengths of this study are its characteristics of an extensive prospective multidisciplinary longitudinal follow-up study design, including repeated follow-up visits, long follow-up period, high number of participants, high participant adherence to the study, data collection from comprehensive national registries covering all participants, and a wide amount of background data of various metrics. To name some limitations, due to our ethnically homogenous population, the generalizability of our findings may be limited to Caucasian/white European subjects. In addition, some level lost to follow-up occurs in all long-term cohort studies. However, comparison of those who had complete ICH data in 1986 against those who did not, demonstrated that they did not differ from cancer incidence, physical activity or smoking habits, which may allow us to assume that participant selection criteria did neither affect the main results (i.e., PA associates with incident cancer) in this study. Reporting of children’s activity and eating behavioral patterns by parents may also be subject to cause some errors and systemic bias but most probably, it can be assumed to be occurred in the same manner in both cancer and non-cancer groups. Furthermore, our cohort still represents rather young individuals, which simultaneously causes a rather low incidence of cancer in general, and the incidence typically increases with age. Therefore, analyses of the association of PA with various cancer subtypes are not yet feasible in this study due to the low number of subtype-specific cancers. This further emphasizes the importance of future continual cohort studies when participants have achieved more age. In the subsequent studies, dissection of the association of physical activity with different cancer subtypes is also highly warranted. Yet we have used previously defined age limits for the youth throughout this study, in which upper age limit being set up to 24 years (34), it is good to recognize that also lower age limits (e.g. range up to 18-21 years) have been previously used for the definition of youth. Although, it has recently been published AHA’s updated recommendation of ‘Life’s essential 8’ (48), also adding sleep as a new subcomponent of healthy life behavior into previous ‘Life’s simple 7’, we had not measured sleep during early surveys of our follow-up and could not thus implement that measure into this study. In this study we could not directly utilize moderate-to-vigorous-intensity physical activity (MVPA) (49), also referred to WHO or US physical activity guidelines, as we have not measure it from the beginning of the cohort follow-up. However, we could generate and estimate MVPA afterwards based by the primary collected PA questionnaire data. Moreover, PAI which we used in our study, has also been previously used measure of PA, and also validated against objective measures (30). In addition, we cannot directly predict causality based on associations found in epidemiological studies due to residual confounding or reverse causality factors, yet we included many confounders simultaneously in the association analysis. However, recent two-sample Mendelian randomization analyses employing summary genetic data results supported a potentially causal relationship between higher physical activity levels and lower risks of breast cancer and colorectal cancer (50), suggesting that PA is a potential primary prevention measure for these cancers.

Conclusion

In this study, we demonstrate an independent association between physical activity in youth and decreased cancer incidence in adulthood, after over 30 years of follow-up. Therefore, although the benefits may not be imminent until decades, initiation of preventive measures and investment in a healthier lifestyle beginning from childhood is essential to promote both heart-healthy and cancer-free life.

Data availability

The dataset supporting the conclusions of this article were obtained from the Cardiovascular Risk In Young Finns study (YFS) after submission and approval of our study plan by the YFS coordinators. The YFS dataset comprises health related participant data and their use is therefore restricted under the regulations on professional secrecy (Act on the Openness of Government Activities, 612/1999) and on sensitive personal data (Personal Data Act, 523/1999, implementing the EU data protection directive 95/46/EC). Due to these legal restrictions, the data from this study cannot be stored in public repositories or otherwise made publicly available.

However, data access may be permitted on a case by case basis upon request only. Data sharing outside the group is done in collaboration with YFS group and requires a data-sharing agreement. Investigators can submit an expression of interest to the chairman of the publication committee (Prof Mika Kähönen, Tampere University, Finland).