Overweight/Obesity and Ancestral Environmental Smoking Exposures: A Summary of Results from the ALSPAC Cohort

Abstract

There is considerable discussion concerning the recent increase in the prevalence of overweight/obesity in children and adults. Although it is assumed that current diet and sedentary behavior are key contributors, these factors do not seem to be the only characteristics responsible. In this paper we summarize the findings we have obtained when assessing whether exposures in previous generations may have played a part in this change over time. In particular, we show that ancestral smoking may be an important contributor. We used data collected from parents and grandparents by the Avon Longitudinal Study of Parents and Children (ALSPAC), which has followed children born in 1991-1992 to women resident in south-west England. We have shown that ancestral smoking characteristics were associated with fetal growth and with increased measures of adiposity in their children and grandchildren. Here we describe the detailed findings of the ancestral exposure to cigarette smoking of ancestors at various time points using ALSPAC data and indicate the support for the findings in other cohorts. Since body mass index (BMI) can be a measure of lean (muscle) mass as well as fat mass, we concentrate on associations with body composition from dual-energy x-ray absorptiometry (DXA). Few birth cohorts have collected data on smoking of individuals in the male line and few have used details of fat, bone, and lean mass. Findings concerning grandmaternal smoking in pregnancy and pre-pubertal smoking of male ancestors were nevertheless replicated. We consider the likelihood of epigenetic explanations for these findings.

Background

There is considerable evidence in many countries including the UK [1] and the US [2-5] that the body mass index (BMI) of the population has increased over time. Interestingly, this increase in BMI is consistent across various European cohorts, with very little difference between them in the upward age-related trajectories between countries [6]. As well as increases in adult BMI, there is considerable evidence that such trends start in childhood, and that the increase over time is particularly apparent in developed countries and in urban areas [5]. Worryingly, at the population level, any increased prevalence of obesity has long-term consequences on poor adult health, especially in regard to increased rates of diabetes, coronary heart disease [7], and mortality in general [8]. Although the time trends with BMI are striking, there is considerable evidence that BMI is not the most appropriate way to measure body fat. A high BMI can occur, for example, in individuals who are very muscular (with high levels of lean mass) [9]. A more appropriate measure of adiposity is therefore an estimate of fat mass.

In order to address the question as to whether there are ancestral environmental exposures that might explain a proportion of these upward trends in BMI, we have used the Avon Longitudinal Study of Parents and Children (ALSPAC) since it has collected environmental data on exposures to parents and grandparents [10,11] as well as measuring fat and lean mass of the offspring over time.

It is recognized that a tendency to overweight or obesity runs in families; but a purely genetic effect cannot account for the marked rise in prevalence over time, and despite over a decade of genetic research there is substantial missing heritability of BMI with only 40-50% of variation explained by genetic variants [12]. Logically, therefore, environmental features must play a part, although interactions between genes and the environment cannot be ruled out. Changes in diet have been considerable over the years and are assumed to play a major part in the etiology of obesity. Here we investigate whether other features experienced prior to conception should also be considered as contributors to the obesity epidemic, using previous results obtained from the ALSPAC cohort.

There is increasing evidence from animal experiments that a chemical insult to an ancestor can have consequences that will be inherited down the generations; for example, exposure to DDT in one generation of rats has been shown to be related to obesity three generations later – the associations being found down both the male and female lines [13]. Similar inter/trans-generational associations have been noted in humans from observational studies in northern Sweden [14]. We have included exposures to previous generations in this review of ALSPAC and other cohort studies.

Aims and Design of this Paper

The Aims of this paper are three-fold: (i) To summarize the data previously published using the ALSPAC data addressing the question as to which ancestral exposures pre-conception are associated with fat mass or BMI or other measures of adiposity in members of the cohort, and whether any such exposures have changed over time; (ii) to determine whether there are other studies in the literature that either support or conflict with the findings; (iii) to consider whether the findings could have an epigenetic explanation.

As a starting point we discuss the results from an exposome hypothesis-free analysis we used to investigate which, if any, of 354 recorded features experienced by parents or grandparents prior to the conception of the study children were independently associated with their fat mass at 24 years of age [15], and whether any of the environmental features identified exhibited trends in prevalence over time.

Materials and Methods

The ALSPAC Cohort

ALSPAC is a cohort of children born 1991-1992 to women who were resident in a defined area (Avon) in south-west England with an expected date of delivery between 1^st April 1991 and 31^st December 1992 [16-18]. The major aim of the study was to identify features of the environment that, together with genetics, may be responsible for various common outcomes [19]. Details of environmental exposures to the parents, grandparents, and great-grandparents were collected from the parents at various time points [10]. Exposures to grandparents and great-grandparents considered in this paper are outlined in the supplementary material attached to [11]. Ethical approval for the study was obtained from the ALSPAC Law and Ethics committee and local research ethics committees (NHS Haydock REC: 10/H1010/70).

The Anthropometric Outcomes Considered

Shortly after birth the offspring were weighed and their lengths and head circumferences measured under standardized conditions. All offspring were invited to in-person examinations from the age of 7 onwards. Examinations included anthropometric measurements including height and weight, from which BMI was calculated. Dual-energy x-ray absorptiometry (DXA) examinations were undertaken at ages 9, 11, 13, 15, 17, and 24 to record estimates of fat and lean mass (see [14] for methodology).

Results of Exposome Analysis

Data collection on exposures pre-conception used frequent structured self-completion questionnaires to each parent, starting in pregnancy. A total of 354 variables were available for the exposome analysis of characteristics that were experienced by the parents and grandparents prior to conception of the study offspring (see Supplementary Material to [14]). Initial analyses selected the unadjusted associations at P<0.05 of the 354 variables with mean fat mass; then, using a time/age structured series of backwards stepwise multiple regressions, the data were reduced until they were still independently associated with fat mass. The analyses were undertaken separately for the factors that were on the maternal and paternal lines (see Figure 1).

Figure 1.

The nomenclature used: PGF (paternal grandfather), and PGM (paternal grandmother) constitute the male (paternal) line; MGF (maternal grandfather), and MGM (maternal grandmother) constitute the female (maternal) line.

Independent Associations of Characteristics in the Maternal Line

The stepwise regression of factors associated with the 24-year-old’s fat mass resulted in the identification of eight independent characteristics on the female line (Table 1a). Of these, three factors that are known to have changed over time were the MGM’s education level, her early age at menarche and mothers’ smoking regularly by the age of 16 (ie, in adolescence). All the factors were adjusted for the year of birth of the MGF, thus partially allowing for cohort effects. The educational level of the MGM cannot, of itself, cause obesity in the grandchild directly, although indirect effects of consequent behavior cannot be ruled out. Theoretically the association with the mother’s early adolescent smoking could be a strong possible contender for mirroring the time trends with obesity in her offspring.

Table 1a. The eight independent features on the maternal line that were associated with offsprings’ fat mass at age 24 (n = 2100; R² = 3.96%); abstracted from [15] .

Variable	Reference	Effect size AMD (95% CI) Kg
MGM’s education level	Lowest level	1.98 (1.11, 2.86)
MGF’s year of birth	Per year	0.10 (0.05, 0.15)
Mother was delivered at 42+ weeks	Gestation <42 weeks	2.27 (0.07, 4.47)
Mother badly scalded <6y	Not scalded <6y	5.42 (1.40, 9.43)
MGF living with mother 6-11y	MGF not living with M	4.46 (1.78, 7.14)
Mother’s age at menarche <12y	M’s menarche >11y	1.74 (0.59, 2.88)
M smoked regularly aged <16y	Did not smoke <16y	2.11 (0.61, 3.61)
M’s parent had a major accident 12-16y	Did not have such an accident	2.79 (0.57, 5.01)

M = mother; MGF = maternal grandfather; MGM = maternal grandmother; AMD = adjusted mean difference.

Independent Associations of Characteristics in the Paternal Line

The final model of independent exposures to the male line that were associated with the 24-year-old’s fat mass consisted of only five factors (Table 1b) – two of which involve smoking (father started smoking regularly pre-puberty (<11); paternal grandmother (PGM) smoked during pregnancy). In addition, father’s social class, based on his occupation will have increased (ie, become less manual) over time. The other variables (father having had a head injury pre-puberty (between ages 6 and 11); and the father being often absent from secondary school (age 11+)), are unlikely to have changed in frequency over time. Consequently, the smoking variables, reflecting the changes in smoking habits over time (Table 2), and increasing affluence (consequent upon changes in social class from manual to non-manual occupations) and lifestyle are potential candidates for contributing to the increasing rates of obesity over time. Henceforth we concentrate on assessing the possible contributions made by ancestral smoking to the adiposity of the fetus, the child, adolescent, and young adult.

Table 1b. The five independent features on the paternal line that were associated with offsprings’ fat mass at age 24 (n = 1975; R² = 6.33%); abstracted from [15] .

Variable	Reference	Effect size AMD (95% CI) Kg
PGF’s social classa	Per class	0.64 (0.28, 1.01)
PGM smoked in pregnancy	Did not smoke in pregnancy	3.21 (1.34, 5.08)
F had head injury aged 6-11y	No head injury this age	2.06 (0.63, 3.49)
F started smoking <11y	Did not start smoking <11y	11.22 (5.23, 17.22)
F often absent from school aged 11+	Not often absent from school	2.80 (1.08, 4.52)

F = Father; PGM = paternal grandmother; PGF = paternal grandfather; AMD = adjusted mean difference. ^aThe higher the social class, the more professional the occupation.

Table 2. The proportion of ALSPAC grandmothers smoking in pregnancy according to their year of birth (in brackets are the numbers smoking in pregnancy) .

Year of birth	MGM	PGM
<1920s	21.3% (59)	20.4% (68)
1920s	22.6% (430)	21.0% (260)
1930s	20.1% (787)	19.3% (381)
1940s	31.4% (877)	29.5% (268)
>1949	45.4% (167)	53.6% (37)
Ptrend	< 0.001	< 0.001

MGM = Maternal grandmother; PGM = paternal grandmother

Summary of Possible Factors Associated with Increasing Time Trends

Of the 354 variables associated with fat mass at age 24 years, 172 and 182 concerned exposures to the maternal and paternal lines, respectively. Reduction revealed that, of these variables, only eight and five of those on the maternal and paternal lines, respectively, remained independently associated with fat mass (Table 1a and 1b); three of the 13 were related to cigarette smoking: paternal grandmother smoking in pregnancy, mother smoking regularly in adolescence and father starting to smoke pre-puberty. These factors remained after adjusting for the other independent variables remaining after backwards stepwise regression. We did not take account of multiple testing by restricting the size of P-values of interest in order to avoid possible Type 1 and Type 2 errors.

The results relating to fat mass at 24-years-of-age raise the question as to whether associations between historical pre-conception smoking experiences were associated with adiposity at earlier ages. The proportion of the population smoking in the UK changed dramatically over the years [20]. For example, in the period 1948-1952, in the age group 25-34, 80% of men and 53% of women were current smokers, but by 1998 the prevalence of smoking had dropped to 39% of men and 33% of women [21]. Much less is known about the change over time in mothers smoking in pregnancy. Retrospective data are, however, available from ALSPAC: Table 2 demonstrates the proportions of the ALSPAC grandmothers who had smoked during pregnancy according to their years of birth. It can be seen that there was a dramatic increase in smoking in pregnancy among women born during and soon after the second World War (the 1940s) and thereafter. For grandfathers, the proportion who had started smoking regularly before age 13 almost doubled between those born prior to 1930 (1.5%) and those born later (2.5%). Similarly, for ALSPAC fathers, the proportion who started smoking before age 13 varied from 2.0% of those born in the 1940s to 7.1% of those born in the 1970s (data not shown).

Henceforth in this paper we therefore describe further analyses of ALSPAC children investigating the extent to which pre-conception ancestral smoking variables were associated with measures of obesity/overweight in the children/grandchildren.

Results of Specific Exposures

Birthweight and grandmother smoking in pregnancy: A study considering measurements at birth in ALSPAC [22] showed that, although babies born to mothers who smoked during pregnancy are twice as likely to be of low birthweight than those born to non-smokers (OR 2.00; 95% CI 1.77, 2.26) [23], if the maternal grandmother (MGM) smoked in the pregnancy resulting in the birth of the mother and the mother did not smoke (MGM+M-), then her grandchild showed an increased birthweight, which was confined to grandsons (adjusted mean birthweight +61 (95 %CI +30, +92) g) but not granddaughters (+14 (-15, +42) g); similar increases in grandsons compared with granddaughters were found for birth length and BMI at birth [22]. There was no association between the smoking of the paternal grandmother (PGM) during pregnancy and birth measurements of the grandchild.

Publications which used other datasets assessing associations between fetal growth and grandmother smoking during pregnancy were mainly concentrating on the MGM, not the PGM; nor did they consider the sexes separately. For example: (a) Rumrich [24] used the MATEX Finnish birth cohort and showed that for MGM+M- there was a decreased risk of the grandchild being small for gestational age (SGA = birthweight <10^th centile) (AOR: 0.83; 95% CI 0.72, 0.94); (b) in Brazil, Magalhaes [25] found no association at P<0.05 for MGM+M- when compared with MGM-M-(Mean difference -58(95% CI -144, +61) g); (c) in the US, Ding [26] showed that if the MGM smoked throughout pregnancy, after allowing for the mother’s smoking during pregnancy, the grandchild’s mean birthweight was 76g heavier than expected (95% CI +18, +134) g. (d) Also in the US, Rillamas-Sun [27] reported on the grandchildren of two cohorts of grandmothers: those born 1904-1928 and 1929-1945 in Michigan – the grandchildren of MGM+M+ in the later cohort were substantially heavier than those MGM-M-: AMD = +346 (+64, +628) g. This excess was not apparent in the earlier cohort.

Thus, like data from ALSPAC, data from Finland and the US comparing MGM+M- with MGM-M- showed an increased birthweight and reduced chance of being SGA, although a similar association was not found in Brazil or the early Michigan cohort.

Parental and grandparental smoking and adiposity in the offspring: As noted above, the ALSPAC exposome analyses identified paternal onset of smoking before puberty (<11 years) and maternal smoking in adolescence (age 11-16) as independent factors associated with the offspring’s fat mass at age 24 [15]. An earlier ALSPAC publication [28] had shown that if the father had started smoking pre-puberty, the adjusted mean differences in BMI, waist circumference and total fat mass of his sons increased with age, being markedly greater from age 13 onwards. There was no such association with daughters’ measurements prior to age 17, but at 17, the daughters had an increased fat mass (AMD = 5.75 95% CI 1.25, 10.2) Kg, although the difference was far greater for sons (AMD = 10.6; 95% CI 5.40, 15.9) Kg.

We also found associations between adolescent (13-16 years) smoking of the paternal grandfathers (PGFs) and the adjusted fat mass of their grandchildren, but no associations with the grandchildren’s lean mass. Grandchildren of the PGF at age 17 had an average excess fat mass of +1.65 (95% CI +0.04, +3.26) Kg, and at age 24 an average excess of +1.55 (95% CI -0.27, +3.38) Kg. Adolescent smoking by the maternal grandfather (MGF) showed similar, but weaker, associations: at 17 years an average excess fat mass of +1.02 Kg (95% CI -0.20, +2.25) Kg, and at 24 an average excess of +1.28 (95% CI -0.11, +2.66) Kg. There were no pronounced differences between the sexes of the children [29].

Elsewhere the following assessments had been made in regard to fathers’ smoking cigarettes: (a) in Northern Norway, the HUNT cohort showed that offspring of fathers who started smoking <11 years had an increased BMI at 12-19 years (+0.97 (+0.06, +1.87)) Kg/m², especially for their daughters (+1.50 (+0.00, +3.00)) Kg/m² but not their sons [30]. (b) Although not exactly the identical measure, the children in the RHINESSA cohort whose fathers smoked pre-conception had offspring with a higher BMI: +0.55 (+0.17, +0.93) Kg/m² [31]. (c) In Hong Kong, individual offspring of women who did not smoke during pregnancy in the Chinese Birth Cohort had a higher BMI at ages 7 and 11 years if their father had smoked during or after pregnancy (childhood height showed no such association at these ages) [32]. (d) In the Irish LIFEWAYS study, the paternal grandmothers’ smoking status (ever by grandchild’s mid-pregnancy) was associated with increased adiposity of the 9-year-old grandchild, measured as a high waist circumference: AOR = 3.29 (95% CI 1.29, 8.37), an association that was particularly associated with the granddaughters: AOR = 3.44 (1.11, 10.69) [33].

Thus, the link found by ALSPAC of increased fat mass if the father started smoking <11 years was also found in the HUNT cohort, and other cohorts showed that paternal smoking prior to the child’s conception was associated with later adiposity.

Grandmother Smoking in Pregnancy and Childhood Adiposity

Previous analyses of ALSPAC anthropometry had shown that if the PGM had, but the study mother had not, smoked in pregnancy (PGM+M-), the granddaughters were taller and both sexes/genders had greater bone and lean mass. However, if the MGM had smoked prenatally but the mother had not (MGM+M-), the grandsons became heavier than expected with increasing age—an association that was particularly due to lean rather than fat mass, reflected in increased strength and fitness. When both the MGM and the mother had smoked (MGM+M+) girls had reduced height, weight, and fat/lean/bone mass when compared with girls born to smoking mothers whose own mothers had not smoked (MGM-M+) [34].

Only a few other longitudinal studies have considered grandmothers smoking in pregnancy and grandchild’s anthropometry with the following results: (a) In the US, using the GUTS2 cohort, Dougan [35] showed that if the MGM had smoked in pregnancy the AOR of being overweight or obese at ages 12 and 17 in granddaughters was slightly greater than in grandsons: AOR = 1.21 (95% CI 0.74, 1.98) v AOR = 1.07 (0.65, 1.77). (b) Using the same cohort, Ding [26] showed that compared to grandchildren of non-smoking women, grandchildren of women who smoked more than 14 cigarettes per day throughout pregnancy (MGM+ v MGM-) were 18% (95% CI: 4%, 34%) more likely to become overweight; their mean BMI was 0.45 kg/m² (95% CI: 0.14, 0.75 kg/m²) higher throughout adolescence and young adulthood than that of grandchildren of non-smoking mothers. (c) The only study to have considered the smoking in pregnancy of the PGM was the Lifeways study in Ireland: the authors showed that the paternal grandmothers’ smoking was positively associated with their 9-year-old grandchild’s overweight/obesity status based on waist circumference (AOR: 3.29, 1.29, 8.37) cm, and especially with that of her granddaughter AOR: 3.44 (1.11, 10.69) cm. These associations remained when restricting analyses to children with biological PGMs (overall: 3.22 (1.25, 8.29)) cm; granddaughter: AOR: 3.55 (1.13, 11.15) cm [33].

Thus, although there are cohorts that show, like ALSPAC, that the grandchildren of MGMs who smoked in pregnancy were more likely to be heavier at birth, there were no studies that used fat mass; similarly, there is no confirmatory evidence of the ALSPAC findings of lean mass since none of the studies that had data of grandmothers smoking in pregnancy had information on lean mass in their grandchildren.

Discussion

In this paper we have described the findings from ALSPAC concerning the smoking history of parents and grandparents prior to the conception of their children/grandchildren and searched the literature for confirmatory findings in regard to the associations with their anthropometry. The two ALSPAC results for which there is confirmatory evidence concern: (i) the association between the paternal grandmother smoking in pregnancy and increased birthweight in the grandchild; this was found in three studies in high-income countries, but not in an early cohort in the US and not in Brazil); (ii) father starting smoking prior to puberty and increased body weight in his offspring (measured as fat mass in ALSPAC and waist circumference in the Lifeways study). None of the sex differences found in ALSPAC were shown in these other studies.

It is disappointing that there are so few studies of intergenerational findings in other datasets. This has resulted in reduced possibilities of being able to confirm our findings of associations between the child’s or grandchild’s adiposity and ancestral smoking habits. The exception has been in regard to the child’s fetal growth, where the MGMs’ smoking in childhood was confirmed to be associated with increased birth measures in several but not all datasets. Lack of such corroboration was found in cohorts born in a low- or middle-income country (Brazil), and in a cohort of births in the US between 1904 and 1928. Thus, it would appear that the findings of increased birthweight when the MGM smoked in pregnancy was most likely to occur in populations with more economic development. It should also be remembered that there are over 1000 different chemicals in cigarettes – the distribution of which is likely to change over time and place.

Some authors have suggested that, if there are such inter/trans-generational effects of the environment, they are more likely to involve nutrition (based on the Overkalix findings). However, Rogers [36] pointed out that the mother smoking during pregnancy has very similar results to the mother being undernourished during pregnancy with lower birthweight, but subsequent increased BMI. Thus, mechanisms may be very similar (at least within one generation).

It is well documented that cigarette smoke includes a large number of toxic chemicals. If our observations are causal, then any one of them may be responsible for the results we have shown. Among the possibilities is DDT, an insecticide first used in 1939 and known to be found in breast milk from smoking mothers [37], and for which there is substantial animal experimental evidence of associations between exposure in one generation and obesity three generations later [12].

Possible Epigenetic Explanation

The idea that the effects of ancestral exposures can be passed down transgenerationally via epigenetic mechanisms remains controversial in mammals [38]. However, there is increasing evidence that there are molecular mechanisms which may enable such transmission, including DNA methylation (DNAm), histone modifications, 3D genome organization, and small non-coding RNAs [38,39]. DNAm is the epigenetic mark most often measured in human epidemiological studies due to its relative stability, and the ease of measurement at large scale with array technologies; thus, it is the epigenetic mark we focus on here. Possible mechanisms of transgenerational transmission include regions which are known to escape the waves of de- and re-methylation in the germ cell and early embryo [40], and the recapitulation of methylation states via molecules such as transcription factors [39]. The most compelling evidence to date in mammals is a demonstration in mice of transgenerational transmission of an edited epigenetic alteration and associated obesity phenotype across four generations [41]; this was by direct DNAm modification rather than via an exposure [42], and DNAm was erased in parental germs cells but re-established in the offspring, illustrating the recapitulation of methylation states via other molecular mechanisms.

Human studies of transgenerational epigenetic inheritance are scarce. None so far have demonstrated the inheritance of epigenetic marks between generations. However, some have identified associations between ancestral exposures and epigenetic patterns in descendants. For example, studies have reported associations that have illustrated the possibility of an association between DNA methylation sites in children and the smoking behavior of their fathers in adolescence [43]; and between DNAm sites in grandchildren, and whether their grandmothers smoked during pregnancy [44]. Two additional studies have found DNAm differences in the grandchildren of women exposed to lead [45] and violence [46] during pregnancy. However, no study as yet has managed to obtain biological samples from all three generations to check the consistency of DNAm changes across all generations involved. It is noteworthy that the study by Watkins was based on peripheral blood results from the ALSPAC cohort and used data with discovery and replication datasets of up to 1225 and 708 individuals, respectively (for the maternal line aged 15-17 years) and tested replication in the same individuals at birth and 7 years. Those analyses showed, for the first time, that DNAm at a small number of loci in cord blood was associated with grandmaternal smoking in humans. In adolescent grandchildren there were suggestive associations in regions of the genome which were hypothesized a priori to be involved in transgenerational transmission – for example there were sex-specific associations at two sites on the X chromosome and one in an imprinting control region. All were within transcription factor binding sites [44].

Strengths and Difficulties

There are several strengths of the ALSPAC results. The analyses of ancestral exposures to cigarette smoke were undertaken taking account of the pivotal findings from the Overkalix study in northern Sweden, where exposure to famine and/or glut in childhood were shown to have long-term associations with outcomes, including mortality [47]. The particular exposures were specific to certain periods of the ancestors’ lives – especially to the pre-puberty period (also described by the authors as the slow growth period). It is of note that there were differences between the male and female lines, such that the paternal line was shown to be the most sensitive to early environmental changes.

No publications to date, apart from our own, have considered associations of birth weight with smoking in pregnancy of the PGM, and none have determined whether associations differed according to the sex of the grandchild. However, the ALSPAC findings of an increase in birthweight if the MGM smoked and the mother did not has been replicated by two other studies [24,26] as well as by one cohort of a study that demonstrated this association for grandmothers born 1929-1945, but not for an earlier birth cohort [27].

Other strengths concern: (a) the fact that ALSPAC is based on residents of a geographic population, and not on specific characteristics, such as college students; (b) the data on ancestors were collected from the study parents without them knowing what specific research questions were being considered; (c) from the age of 9 onwards, the study cohort was examined using a Dexa machine which was able to distinguish fat and lean mass, thus enabling more specific measures than BMI into early adulthood; (d) information is available to link the data to the DNA methylation results on the study children.

The disadvantages of this study lie in the fact that (i) few independent longitudinal studies have collected appropriate details so that our results can be validated – this is particularly true of exposures to members of the male line – particularly the paternal grandparents, and of the anthropometric components of body weight, such as fat and lean mass; (ii) the study is concerned almost wholly with individuals of a White European ethnic group which was representative of the Avon population in the 1990s – thus the results cannot be extrapolated to Multiracial heterogeneous populations; (iii) much of the data on the ancestors was collected from the study parents, and has all the problems of retrospective recall – however, the paper questionnaires used were designed to be completed in the individual’s own home, thus allowing the participants being able to contact relatives or family friends for details.

Conclusions

Using data from a single pre-birth cohort we have been able to demonstrate confirmatory evidence for some, but not all of our grandparental smoking associations with adiposity in the grandchild. However, there is a lack of data concerning ancestral cigarette smoking during childhood and during pregnancy, particularly in the paternal line, for confirmatory analyses. Intergenerational or transgenerational associations in humans should be investigated further, as they may be particularly important in the future understanding of chronic overweight/obesity and all the consequential health problems.

Glossary

ALSPAC

Avon Longitudinal Study of Parents and Children

AOR

Adjusted odds ratio

AMD

Adjusted mean difference

BMI

Body Mass Index

CI

Confidence interval

DNAm

DNA methylation

DXA

dual-energy x-ray absorptiometry

F

Father

M

Mother

MD

Mean difference

MGF

Maternal Grandfather

MGM

Maternal Grandmother

MGM+

Maternal grandmother smoked in pregnancy

MGM+M-

MGM smoked in pregnancy, mother did not

PGF

Paternal Grandfather

PGM

Paternal Grandmother

SGA

small for gestational age

Author Contributions

JG and SW wrote the first draft; all authors (JG, SW, KN: https://orcid.org/0000-0002-0602-1983, MS: https://orcid.org/0000-0002-2715-9930, and YI-C: https://orcid.org/0000-0002-9965-9133) contributed to editing and rewriting.

Funding Statement

The UK Medical Research Council and Wellcome (Grant ref: 217065/Z/19/Z) and the University of Bristol currently provide core support for ALSPAC. This publication is the work of the authors and JG will serve as guarantor for the contents of this paper. Specific funding contributing to this focused review was provided by the John Templeton Foundation (grant no. 91816).