Introduction
Early life growth-restriction is a delay in normal growth during development, Reference Hui and Challis1–Reference Zozaya, Diaz and de Pipaón3 and is typically a result of placental insufficiency, low-food availability to the mother or infant, and/or premature birth. Reference Zozaya, Diaz and de Pipaón3,Reference Figueras and Gardosi4 Unfortunately, growth-restriction affects 3–7 percent of all births across the globe. Reference Romo, Carceller and Tobajas5,Reference Black, Victora and Walker6 The large number of children (160 million) experiencing growth-restriction in early life is alarming, as reduced growth in early life is associated with an increased risk for chronic diseases (cardiovascular disease, type II diabetes, and obesity) in adulthood. Reference Barker, Gluckman, Godfrey, Harding, Owens and Robinson7–Reference Shankaran, Das and Bauer11
Researchers have demonstrated that physical activity engagement reduces the incidence of chronic disease. Reference Bauman12–Reference Luke, Dugas, Durazo-Arvizu, Cao and Cooper15 Furthermore, literature has shown a dose-response effect with physical activity, where those who engage in greater amounts of physical activity experience greater reductions in disease risk. Reference Pate16 Physical activity could therefore serve as an effective therapeutic countermeasure to the negative health consequences from early life growth-restriction. Studies have demonstrated that growth-restricted individuals perform less physical activity than normal weight offspring. Andersen et al evaluated 13 Nordic birth cohorts and found that individuals with very low and very high birth weights (< 2.0 kg or > 5.0 kg) were the least physically active in adulthood. Reference Andersen, Angquist and Gamborg17 Recent animal studies published from our laboratory demonstrated similar findings to those of Andersen et al. Specifically, female mice that were growth-restricted in early life (days 1–21 of postnatal life) engaged in significantly less physical activity compared to non-growth-restricted controls. We hypothesize this reduction in physical activity reduces unnecessary caloric expenditure, which would be beneficial in times of low food availability (for additional information, see Reference Andersen, Angquist and Gamborg17–Reference Hales and Barker21 ). Furthermore, Ferguson et al recently found that growth-restricted mice did not experience positive health benefits from a three week voluntary wheel running exercise intervention. Reference Ferguson, Leszczynski, McPeek, Pendergrast, Visker and Triplett22 The lack of adaptations following physical activity, in addition to an already increased risk for chronic disease development, predispose growth-restricted individuals to an alarming predisposition for premature death. However, the results observed in previous animal studies have not been tested in humans and should be studied further before making widespread recommendations.
Human physical activity studies that include low birth weight/growth-restricted individuals and longitudinal reporting of physical activity levels could accurately determine the change over time in response to this early life impairment. The Framingham Offspring Cohort, a derivative of the Framingham Heart Study, was designed to follow children from birth for over 40 years. Dataset measures include birth weight, growth trajectory, and adult physical activity levels among many other outcomes. Collection of this information makes the Framingham Offspring Cohort an ideal dataset to determine the relationship between low birth weight, physical activity engagement, and the relationship between physical activity and chronic disease. Thus, data available through the Framingham Offspring Cohort can be used to examine the previous animal findings observed in our lab and others in a well-controlled human cohort. The purpose of this investigation was to examine the relationship between Framingham Offspring Cohort individuals born low birth weight (LBW, defined as < 2,500 g at birth Reference Cutland, Lackritz and Mallett-Moore23 ), and biomarkers of chronic disease, adult physical activity levels, and chronic disease status at two different time points (exam 5 visit [average age: 50 ± 8], exam 9 visit [average age: 69 ± 8]) in comparison to normal birthweight (NBW) individuals. We hypothesized that individuals born LBW would have elevated biomarkers of chronic disease compared to NBW individuals. We also hypothesized LBW individuals would engage in significantly less physical activity compared to NBW individuals at both exam visits. Finally, we hypothesized that LBW individuals would have higher likelihood to be treated for chronic diseases compared to NBW participants when accounting for physical activity engagement.
Methods
Data acquisition and approval
Institutional Review Board approval was obtained from Michigan State University to ensure participant privacy. Data from the Framingham Offspring Cohort Data were supplied from the Biologic Specimen and Data Repository Information Coordinator Center (BioLINCC). Participants were included in the analysis if both birth and physical activity information were available (for both Cohort visits) and those who reported their birthweight with a moderate level of confidence (a very certain or somewhat certain rating), or birthweight information was obtained directly from birth certificates. If participants had any missing physical activity data, a missing data estimation of means was performed utilizing the other physical activity parameters to estimate the value in SPSS. For example, if one intensity of physical activity collected was missing, the remaining intensities (respectively for each timepoint) were used to estimate the missing value. The final analytic sample consisted of 1,018 participants at exam 5 and 995 at exam 9. There was approximately a 19-year difference between the two visits, and as such the first visit (visit 5) was defined as adulthood (average age: 50 ± 8 years), and information from the second visit (visit 9) as late adulthood (average age: 69 ± 8 years). Additional information on the Framingham Offspring Cohort, including study recruitment, study protocols, and study timeline have been discussed in previous reports. Reference Liu, Meigs and Pittas24–Reference Kannel, Feinleib, McNamara, Garrison and Castelli27
Risk factors and demographic information
Risk factors such as treated for diabetes, hypertension, or dyslipidemia were collected in addition to demographic information (age, sex). Specifically, participants were asked if they were taking medication for these diseases, and if they were, participants were classified as such. The Framingham researchers collected blood samples for calculation of blood triglycerides, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C) and blood glucose at each exam visit (exam 5 [adulthood], exam 9 [late adulthood]) which were used as risk for cardiometabolic disease.
Questionnaires were provided to collect participant birth weight information and the source from which birthweight was obtained, including confidence in the value provided. Participants were excluded if they rated their confidence in the number not very certain or did not know their birthweight. At each visit, participants had their height and weight measured (used to calculate BMI) as well as blood pressures.
Physical activity measurement
During exam 5, physical activity engagement was measured via questionnaire responses collected through interviews. Participants were briefly asked questions such “Number of hours with activities such as standing, walking” (defined as slight activity), or “Number of hours with activities such as housework (vacuum, dust, yard chores, climbing stairs; light sports such as bowling, golf)” (defined as moderate activity). Researchers also collected additional physical activity data including hours spent sedentary and hours spent engaging in heavy activity via interview questions with a Framingham researcher.
During exam 9, physical activity was measured via accelerometers (Actical Accelerometer [model #198-0200-00]) worn on the right hip at mid clavicular line for a total of 7 days during all waking hours. Specific information from accelerometers included time engaging in sedentary activity, time engaging in light intensity of physical activity, time engaging in moderate intensity physical activity, time engaging in vigorous physical activity, time engaging in moderate-to-vigorous physical activity (MVPA), and longest bouts of MVPA, sedentary, light, moderate, and vigorous activity. All physical activity data was calculated by Framingham researchers following established cut points. Reference Crouter and Bassett28,Reference Heil29 Total physical activity engagement was then calculated at each time point separately by combining light, moderate, and heavy physical activity (exam 5), or slight, moderate, and vigorous physical activity (exam 9) into a single outcome. Total physical activity was then ranked into terciles of low, medium, or high physical activity engagement per time point. At exam 5, total physical activity resulted in tercile rankings of 7.30, 9.78, and 12.75 hours of total physical activity for low, medium, and high, respectively. At exam 9, low, medium, and high terciles engaged in 1.35, 1.85, and 2.46 hours of physical activity, respectively. Participants were ranked into their corresponding tercile for further analysis.
Statistics
For the first aim, a two-way ANCOVA was performed to compare the main effects of sex and birthweight status, and the sex by birthweight interaction on each marker of chronic disease (non-fasting blood glucose, diastolic and systolic blood pressure, BMI, triglycerides, and cholesterols) at both time points. Age was the covariate. For the second aim a two-way ANCOVA was run with sex, birthweight status, and the sex by birthweight classification interaction as main effects, physical activity variables (spent engaging in sedentary, light, moderate, and heavy/vigorous activity) as outcomes. Covariates were blood glucose, BMI, age, systolic blood pressure, and total cholesterol. Finally, a binomial logistic regression was performed with birthweight and sex as independent factors, tercile of total physical activity (low, medium, high) as a covariate, and chronic disease status (treated for diabetes, hypertension, dyslipidemia) at both exam visits (exam 5, exam 9) as outcomes. To further clarify the association of physical activity with chronic disease development, a two-way ANCOVA was performed comparing the main effects of sex, birthweight status, and the sex by birthweight interaction on markers of the previously mentioned biomarkers of chronic disease, with age and total physical activity as a covariate at each timepoint. Statistical significance was denoted as a p value of Alpha < 0.05. Results approaching statistical significance (p value of Alpha < 0.05–0.10) are also discussed to highlight differences with potential biological relevance, though we note there is a higher likelihood these results may have occurred by chance than statistically significant results.
Results
Markers of chronic disease
Participant demographic information and chronic disease biomarkers of interest at exams 5 and 9 are presented in Table 1 and Table 2, respectively. At exam 5 (average age 50 ± 8, 64.5% female, 10.4% LBW), there were significant birthweight and sex main effects, where LBW participants had significantly higher levels of blood glucose (difference in estimated means: +6.9 mg/dL, p = 0.001) compared to normal birthweight (NBW) individuals. Males had significantly higher levels of blood glucose (+11.0 mg/dL, p < 0.001), diastolic blood pressure (+4.4 mmHg, p < 0.001), and systolic blood pressure (+8.7 mmHg, p < 0.001) than females. Females had significantly higher HDL-C (+14.0 mg/dL, p < 0.001) and lower LDL-C (−10.9 mg/dL, p = 0.005) compared to males. There were no differences in total cholesterol between any groups.
Demographic information exam 5
Bold and underlined indicate a significant difference from two-way ANCOVA (age used as covariate). *-Significant birthweight main effect (p < 0.05). Δ-Significant sex main effect (p < 0.05). BMI, Body Mass Index; HDL-C, High-Density Lipoprotein cholesterol; LDL-C, Low-Density Lipoprotein cholesterol; SBP, Systolic Blood Pressure; DBP, Diastolic Blood Pressure. Values are presented as means ± standard deviation.
Demographic information exam 9
Bold and underlined indicate a significant difference from two-way ANCOVA (age used as covariate). *-Significant birthweight main effect (p < 0.05). Δ-Significant sex main effect (p < 0.05). Results approaching significance (0.05<p < 0.10) are italicized. BMI, Body Mass Index; HDL-C, High-Density Lipoprotein cholesterol; LDL-C, Low-Density Lipoprotein cholesterol; SBP, Systolic Blood Pressure; DBP, Diastolic Blood Pressure. Values are presented as means ± standard deviation.
At the exam 9 visit (average age 69 ± 8 years old, 63.0% female, 9.9% LBW), LBW participants had significantly higher levels of blood glucose (+7.8 mg/dL, p = 0.002) and resting triglycerides (+14.5 mg/dL, p = 0.023) compared to normal birthweight participants. LBW participants potentially had lower resting levels of HDL-C compared to those born normal birthweight (-3.6 mg/dL, p = 0.085). Males had significantly higher levels of blood glucose (+13.4 mg/dL, p < 0.001) and diastolic blood pressure (+3.2 mmHg, p = 0.001) compared to females. Females, conversely, had significantly higher blood HDL-C (+15.9 mg/dL, p < 0.001), LDL-C (+15.1 mg/dL, p < 0.001), and total cholesterol (+29.7 mg/dL, p < 0.001) compared to males. There were no significant birthweight by sex interactions with markers of chronic disease.
Physical activity engagement
At exam 5, LBW males potentially engaged in greater amounts of slight activity compared to LBW females (+0.55 hours, p = 0.09, Table 3). LBW females potentially spent more time sedentary than NBW females (+0.72 hours) and LBW males (+0.82 hours), though also not statistically significant (p = 0.08). LBW females potentially engaged in less total physical activity than LBW males (−0.76 hours) and normal birthweight males (−0.38 hours)/females (−0.68 hours, p = 0.078). Males spent significantly greater time (+0.40 hours, p = 0.005) doing heavy activity compared to females. However, there were no significant interaction (birthweight by sex) associations for physical activity behavior (slight activity, moderate activity, heavy activity, or total activity) or sedentary behavior.
Physical activity exam 5
Bold and underline indicates a significant difference from two-way ANCOVA (blood glucose, BMI, age, systolic blood pressure, and total cholesterol covariates). Δ-Significant sex main effect (p < 0.05). Results approaching significance (0.05<p < 0.10) are italicized. Blood glucose, BMI, age, systolic blood pressure, and total cholesterol used as covariates in analysis. All data presented as means ± standard deviation.
At exam 9, LBW males spent significantly more time sedentary (data collected via accelerometry) compared to NBW males (+0.31 hours) and LBW females (+0.30 hours, Table 4). There were no other significant interaction associations for light, moderate heavy, or MVPA. There were no significant associations of LBW on any physical activity outcomes. However, females engaged in significantly less moderate physical activity (+0.09 hours, p = 0.003) and total MVPA minutes (+0.10 hours, p = 0.004) compared to males. Though not statistically significant, females potentially engaged in less light activity compared to males (+0.12 hours, p = 0.095).
Physical activity exam 9
Bold and underline indicates a significant difference from two-way ANCOVA (blood glucose, BMI, age, systolic blood pressure, and total cholesterol covariates). Different superscript letters denote significant differences between groups (p < 0.05). Δ-Significant sex main effect (p < 0.05). Results approaching significance (0.05<p < 0.10) are italicized. Values are presented as means ± standard deviation.
Physical activity and health outcomes
Roughly 1.2 percent of Framingham Offspring Cohort participants were treated for diabetes at the exam 5 visit, slightly below the national diabetes prevalence rate of 3.6 percent during this time period (1990–1991). Reference Mokdad, Ford and Bowman30 Only 10.4 percent of participants at the exam 5 visit were treated for hypertension, also lower than the national prevalence of 27 percent for individuals in the 40–59 years of age bracket from 1988–1991. Reference Hajjar and Kotchen31 Finally, 4.7 percent of participants were treated for dyslipidemia during the exam 5 visit, again lower than the national prevalence of elevated triglycerides (31.8% males, 21.6% females) or low HDL-C (29.6% males, 35.3% females) in adults 18 years or older. Reference Moore, Chaudhary and Akinyemiju32
When accounting for the levels of physical activity engagement, there were no significant associations of low birthweight on participant likelihood to be treated for diabetes (p = 0.477, Table 5.) However, females were, potentially, less likely to be treated for diabetes (p = 0.062, Odds Ratio: 0.34 [0.11, 1.05]) compared to males. There were no significant differences in birthweight (p = 0.997) or sex (p = 0.327) for likelihood to be treated for hypertension at exam 5. Females were significantly less likely to be treated for dyslipidemia compared to males (p = 0.021, Odds Ratio: 0.51 [0.28, 0.90]). There were no significant associations for birthweight in likelihood of treatment for dyslipidemia amongst participants (p = 0.333).
Odds ratios for chronic disease development at exam 5
Bold text indicates significant difference from binomial logistic regression. *-Significantly lower odds for females to be treated for disease (p < 0.05) compared to males. NBW, Normal Birthweight; LBW, Low Birthweight; OR, Odds Ratio; CI, Confidence Interval; Ora, Adjusted Odds Ratio for tercile of physical activity.
At the exam 9 timepoint, 10.7 percent of participants were treated for diabetes, well below the national prevalence of 21.8 percent in adults aged 65 or older from 2005–2010. Reference Selvin, Parrinello, Sacks and Coresh33 Additionally, 55.3 percent of participants were treated for hypertension at exam 9, also below the national hypertension prevalence of 70.8–71.6 percent for those over the age of 65 from 1999–2010. Reference Gillespie and Hurvitz34,Reference McDonald, Hertz, Unger and Lustik35 Finally, 50.1 percent of participants were treated for dyslipidemia during the exam 9 visit, slightly below the national prevalence for adults aged 45 and older of 56.8 percent Reference Lu, Wang and Zhou36 and below the national average of dyslipidemia from 1999–2004 for adults over 65 of 60.3 percent. Reference McDonald, Hertz, Unger and Lustik35
Females were less likely to be treated for diabetes (p < 0.001, Odds Ratio: 0.57 [0.44, 0.74], Table 6) when controlling for levels of physical activity. Low birthweight was not significantly associated with the likelihood to be treated for diabetes (p = 0.133) when adjusting for physical activity. Females were significantly less likely to be treated for hypertension at exam 9 (p = 0.021, Odds Ratio: 0.74 [0.57, 0.96]). Birthweight was not a significant predictor for being treated for hypertension (p = 0.549). Lastly, women were less likely to be treated for dyslipidemia (p < 0.001, Odds ratio: 0.57 [0.44, 0.74]). Birthweight was not predictive of likelihood for treatment of dyslipidemia (p = 0.98).
Odds ratios for chronic disease development at exam 9
Bold text indicates significant difference from binomial logistic regression, while italicized text indicates a statistical trend (0.05<p < 0.10). *-Significantly lower odds for females to be treated for disease (p < 0.05) compared to males. Results approaching significance (0.05<p < 0.10) are italicized. NBW, Normal Birthweight; LBW, Low Birthweight; OR, Odds Ratio; CI, Confidence Interval; Ora, Adjusted Odds Ratio for tercile of physical activity.
When age and total physical activity engagement were included as covariates in the analysis of chronic disease biomarkers during exam 5, LBW participants had elevated blood glucose (+6.9 mg/dL, p = 0.001, Table 7) compared to NBW participants. There were no other significant differences in the LBW or LBW by sex interaction groups. Females had greater HDL-C (+14.0 mg/dL, p < 0.001), but lower blood glucose (−11.0 mg/dL, p < 0.001), LDL-C (−10.9 mg/dL, p = 0.005), BMI (−1.4 kg/m2, p = 0.016), DBP (−4.4 mmHg, p < 0.001) and SBP (−8.7 mmHg, p < 0.001) compared to males.
Biomarkers of chronic disease exam 5
Bold and underline indicates a significant difference from two-way ANCOVA (age and total physical activity used as covariates). *-Significant birthweight main effect (p < 0.05). Δ-Significant sex main effect (p < 0.05). BMI, Body Mass Index; HDL-C, High-Density Lipoprotein cholesterol; LDL-C, Low-Density Lipoprotein cholesterol; SBP, Systolic Blood Pressure; DBP, Diastolic Blood Pressure. All data presented as least square means ± standard error.
At the exam 9 timepoint, blood glucose (+7.7 mg/dL, p < 0.001, Table 8) and blood triglycerides (+14.3 mg/dL, p = 0.019) remained significantly elevated in the LBW group compared to NBW participants when total physical activity and age were used as covariates. Females had lower DBP (-3.1 mmHg, p = 0.003) and blood glucose levels (−13.8 mg/dL, p < 0.001) compared to males. Females also had elevated HDL-C (+16.3 mg/dL, p < 0.001), LDL-C (+15.4 mg/dL, p < 0.001), and total cholesterol (+30.2 mg/dL, p < 0.001) compared to males. There were no significant birthweight by sex interactions.
Biomarkers of chronic disease exam 9
Bold and underline indicates a significant difference from two-way ANCOVA (age and total physical activity used as covariates). *-Significant birthweight main effect (p < 0.05). Δ-Significant sex main effect (p < 0.05). BMI-Body Mass Index, HDL-C-High-Density Lipoprotein, LDL-C-Low-Density Lipoprotein, SBP-Systolic Blood Pressure, DBP-Diastolic Blood Pressure. All data presented as least square means ± standard error.
Discussion
Early life growth-restriction is characterized by a variety of growth impairments throughout development, including conditions such as small-for-gestational age, low birthweight, or extrauterine growth-restriction. Reference Hui and Challis1-Reference Zozaya, Diaz and de Pipaón3,Reference Goldenberg and Cliver37,Reference Fenton, Cormack and Goldberg38 Since a brief window of early life growth-restriction can increase the risk for chronic disease in adulthood, Reference Osmond and Barker10,Reference El Hajj, Schneider, Lehnen and Haaf39 it is necessary to examine potential therapeutic interventions and their effectiveness in reducing the disease risk. Physical activity, for example, reduces the risk of developing chronic diseases. Reference Lee, Shiroma and Lobelo14,Reference Paffenbarger, Blair and Lee40 Recent evidence, however, suggests that a brief window of early life growth-restriction can inhibit the positive adaptions to adult physical activity. Reference Ferguson, Leszczynski, McPeek, Pendergrast, Visker and Triplett22 While alarming, the lack of adaptations has been observed in mice experiencing early life growth-restriction and thus human data must be analyzed to confirm the potential maladaptive response. The vast amount of longitudinal data from the Framingham Offspring Cohort makes it an effective data set to analyze the effects of early life growth-restriction (defined in this cohort as born low birthweight) on both physical activity engagement and chronic disease development.
The lower rates of chronic disease treatments in the Framingham Offspring Cohort observed in this analysis were somewhat surprising, as previous analysis of the Framingham Heart Study demonstrated data from the Framingham Community were comparable to the national results observed in the National Health and Nutrition Examination Survey amongst white men. Reference Leaverton, Sorlie and Kleinman41 However, these results were observed in the original Framingham Heart Study Cohort and not their children (as analyzed in this study), and method of disease measurement (including diabetes and cardiovascular disease) were not identical between sample groups. In addition to differences in disease measurement, more recent literature suggests differences between national results observed in NHANES reports and those in the Framingham Heart and Offspring study may also be due to differences in geographical locations of participants and differences in racial distribution of study participants. Reference Preis, Hwang and Coady42 Regardless, trends and risk factor models generated from Framingham Heart Study have good to strong applicability when applied to various populations (c-statistic ranging from 0.63–0.83). Reference D’Agostino, Grundy, Sullivan and Wilson43,Reference Bitton and Gaziano44
Participants in our analysis had a LBW rate of 10.4 percent (106 of 1,018 participants) at exam 5, and 9.9 percent (99 of 995 participants) at exam 9. Though the average BMI of participants was categorized as overweight, LBW participants did not have lower BMI at exam 5 or exam 9 compared to NBW participants. This is surprising, as previous studies demonstrate early life growth-restriction reduces adult weight and mass. Reference Leszczynski, Visker and Ferguson20,Reference Ferguson, Leszczynski, McPeek, Pendergrast, Visker and Triplett22,Reference Ferguson, Monroe and Heredia45,Reference Visker and Ferguson46 However, these findings could be explained by higher rates of obesity and adiposity. For example, Jaquet and colleagues demonstrated intrauterine growth-restricted born adults had significantly higher percent body fat at 25 years of age, despite no differences in BMI. Reference Jaquet, Gaboriau, Czernichow and Levy-Marchal47 Additionally, Barker demonstrated greater obesity rates in individuals born during famine compared to those born without early life growth-stunting. Reference Barker8,Reference Barker and Osmond48,Reference Barker49 As BMI is not the same as body composition (muscle and fat mass), the lack of differences between LBW and NBW participants may be masked. For example, LBW participants at exam 9 may have higher adiposity and subsequent lower muscle mass compared to NBW individuals, resulting in no observed BMI differences.
LBW participants had higher resting blood glucose levels at exam 5, which continued 20 years later at exam 9. Additionally, LBW participants had higher levels of triglycerides at exam 9 and, though not statistically significant, lower levels of high-density lipoproteins. The elevated levels of blood triglycerides and blood glucose in LBW individuals are in agreement with previous research demonstrating higher levels of type II diabetes, metabolic syndrome, and hyperlipidemia in adults who were growth-restricted in early life. Reference Barker8,Reference El Hajj, Schneider, Lehnen and Haaf39,Reference Barker50 The elevated biomarkers are hypothesized to be a result of altered organ growth (kidneys, liver) to spare growth of more important organs (brain, heart). Reference Hales and Barker21,Reference Hales and Barker51,Reference Hales52 Studies have demonstrated the deleterious effects of early life growth-restriction on liver function and structure, including changes in hepatic bioenergetics, Reference Lane, Flozak, Ogata, Bell and Simmons53 hepatic oxidative stress and mitochondrial stress, Reference Oke and Hardy54 and altered hepatic glucose transport, Reference Cianfarani, Agostoni and Bedogni55,Reference Dessì, Ottonello and Fanos56 which could lead to nonalcoholic fatty liver disease and eventually type II diabetes. Reference Anstee, Targher and Day57
Previous studies have shown that growth-restricted individuals engage in significantly less physical activity compared to controls. Reference Andersen, Angquist and Gamborg17,Reference Leszczynski, Visker and Ferguson20,Reference Davies, Smith, May and Ben-Shlomo58 However, the average age of participants in these studies were relatively young (average age for Davies et al 38 ± 7.9, Andersen et al ranging from 14–69), or were limited in the analysis of specific physical activity intensities (e.g., differences in light, moderate, or vigorous activity). Fortunately, physical activity data from the Framingham Offspring Cohort included various physical activity intensities, which allowed for specific determination of differences in physical activity behavior. However, it is important to note the differences in the method of physical activity data collection between the different time points. At exam 5, physical activity data were collected via interview, with activities being coded into metabolic equivalent values then categorized as slight, moderate, or heavy activity. Exam 9 physical activity data was collected via accelerometry. The contrast in physical activity collection method, in addition to age-related declines in physical activity, Reference Troiano, Berrigan, Dodd, Masse, Tilert and McDowell59,Reference Buchman, Wilson, Yu, James, Boyle and Bennett60 helps explain the considerable differences in physical activity engagement between the two time points.
Though not statistically significant, LBW females spent more time sedentary and less time engaging in slight and total activity during the exam 5 visit. Conversely, LBW males spent significantly more time sedentary than LBW females and NBW males at exam 9, and no other differences in other physical activity outcomes were present. These results were surprising, as females tend to be more sedentary than males. Reference Gardner and Montgomery61–Reference Hallal, Andersen and Bull63 However, recent literature by Leszczynski and colleagues revealed female postnatally growth-restricted mice engaged in significantly less physical activity (measured via wheel running) in adulthood. Reference Leszczynski, Visker and Ferguson20 Additionally, Baker and colleagues, found a female-specific detriment in adult physical activity engagement and higher rates of obesity in mice experiencing fetal growth-restriction. Reference Baker, Li, Kohorst and Waterland19 Therefore, the results observed at exam 5 (LBW females potentially spending more time sedentary with reduced slight and total physical activity) appear to be consistent with results observed in previous studies. However, the reversal in results at exam 9 (LBW males spending more time sedentary than LBW females and NBW males), in conjunction with higher markers of chronic disease (greater triglycerides and blood glucose), could be indicative of greater disease state progression at older ages in the male population, potentially due to differences in genetics (sex chromosomes) or sex-specific behaviors. Reference Bupp64 For example, Bupp describes that both sexes experience declines in the immune system due to aging, though males experience this decline in immune function sooner and at a faster pace compared to women. Reference Bupp64 It is also possible, given greater occupational physical demand of men in the Framingham Offspring Cohort, Reference Felson, Hannan and Naimark65 differences in physical activity during exam 5 may be due to greater work activity that was not present upon retirement during the exam 9 timepoint. Future research should attempt to elucidate the sex differences observed in this study.
Ferguson et al recently demonstrated growth-restricted mice do not experience positive health adaptations following 3 weeks of voluntary wheel running. Reference Ferguson, Leszczynski, McPeek, Pendergrast, Visker and Triplett22 As growth-restricted individuals already experience a significantly increased risk for chronic disease development, Reference Barker8,Reference El Hajj, Schneider, Lehnen and Haaf39,Reference Paneth and Susser66 the lack of positive adaptations is alarming. Therefore, to confirm the findings demonstrated in animal models, a multinomial logistic regression was performed on chronic disease outcomes (treated for diabetes, treated for hypertension, treated for dyslipidemia) with LBW status and sex as main effects and tercile of physical activity as a covariate within the model.
The results suggest that LBW individuals are not at increased risk for chronic disease treatment (diabetes, hypertension, dyslipidemia) compared to NBW participants. The lack of differences in odds ratios was present both in the unadjusted model and when adjusted for physical activity. It is important to note, however, that while the likelihood of being treated was not different between birthweight groups, the biomarkers of chronic disease remained significantly elevated in the LBW group when total physical activity was used as a covariate. The elevated biomarkers of chronic disease in the LBW group suggest that LBW participants may not experience health benefits following physical activity engagement and the detrimental effects of early life growth-restriction may not be attenuated in adulthood. Ferguson and colleagues observed similar findings in their animal model where female growth-restricted mice had reduced left ventricular volume and left ventricular area following three weeks of voluntary wheel running. Reference Ferguson, Leszczynski, McPeek, Pendergrast, Visker and Triplett22 While in agreement with previous animal studies, there are limitations in comparing the results observed in this study and previous animal work. For example, participant information on the potential effects of extrauterine growth-restriction was not collected. Individuals in the LBW group may have had a wide range of postnatal outcomes/growth following the initial LBW classification, which may have altered their adult health and phenotype. Additionally, while a lack of adaptions following physical activity was observed in mice experiencing postnatal growth-restriction, Reference Ferguson, Leszczynski, McPeek, Pendergrast, Visker and Triplett22 the growth-faltering experiencing during this developmental window corresponds to the third trimester of gestation in humans and early postnatal life. Reference Ferguson, Monroe and Heredia45 Information on in utero conditions were not available (small-for-gestational age, premature birth) in the Framingham Offspring Cohort, and thus participants in the Framingham Offspring Cohort may not have experienced identical epigenetic programing to previous results observed in previous animal studies. To address these concerns, future longitudinal studies must be conducted to track not only birthweight and intrauterine conditions, but also early postnatal/childhood growth to measure the degree of growth-faltering and fully illuminate the effects of early life growth-restriction on adult outcomes.
Females were 43 percent less likely than males to be treated for diabetes during the exam 5 visit and were 66, 26, and 43 percent less likely to be treated for diabetes, hypertension, and dyslipidemia, respectively, at the exam 9 timepoint. The differences in chronic disease development between males and females are in accordance with previous literature demonstrating that men have higher prevalence of diabetes, Reference Peters, Muntner and Woodward67,Reference Tramunt, Smati and Grandgeorge68 hyperlipidemia, Reference Hill and Bordoni69 and cardiovascular disease, Reference Mosca, Barrett-Connor and Kass Wenger70 with women experiencing delayed development of cardiovascular disease compared to men. Reference Maas, Van Der Schouw and Regitz-Zagrosek71 However, these differences may be due to differences in the treatments between the sexes. Specifically, with regards to hypertension, deaths from cardiovascular disease have been progressively rising in the female population despite historically being thought of as a disease more prevalent in males. Reference Maas, Van Der Schouw and Regitz-Zagrosek71,Reference Towfighi, Zheng and Ovbiagele72 Therefore, the difference in likelihood to be treated for hypertension may be a result of lower awareness of cardiovascular risk factors, and as a result lower rates of treatment. Reference Maas, Van Der Schouw and Regitz-Zagrosek71,Reference Gao, Chen, Sun and Deng73 Further research into the differences between the sexes is warranted to further clarify the differences in development of chronic diseases.
Conclusion
LBW individuals in the Framingham Offspring Cohort demonstrated significantly higher rates of chronic disease markers (resting blood glucose, triglycerides) and greater levels of sedentary behavior in later adulthood (average ages 50 and 69). However, LBW participants in the Framingham Offspring Cohort were not more likely to be treated for chronic diseases when controlling for physical activity engagement, though biomarkers of chronic disease remained elevated. The lack of differences in likelihood to be treated for disease, despite elevated biomarkers of chronic disease, should be investigated further in this at-risk population. Additionally, as adulthood physical activity did not reduce the elevated biomarkers of chronic disease, behavioral interventions in the growth-restricted population may need to be implemented prior to adulthood (for example, increasing childhood and adolescence physical activity) to promote lifelong physical activity and ultimately offset the detrimental effects of developmental programing and subsequent development of chronic diseases at later timepoints.
Limitations
Utilization of the Framingham Offspring Cohort allowed for analysis of biological and behavioral data through decades of collection, though there are a few limitations. For example, there is a potential for response bias from the Exam 5 visit as data was collected through self-reporting. Furthermore, the difference in physical activity data collection (interview/questionnaire at exam 5, accelerometer at exam 9) means direct comparison of physical activity behavior between the two time points was not possible. While we have focused primarily on the role of physical activity engagement in the prevention of chronic diseases, we must also note there is potential for reverse causality where those with chronic diseases are unable to engage in physical activity and thus have lower levels of physical activity. The older demographic in this study allowed for unique analysis of the effects of birthweight on an aging population. However, the high amount of light/slight activity may limit the overall effects of physical activity. While recommendations for older adults are the same for younger adults (150 minutes per week of moderate intensity exercise), light exercise still provides benefits to the aging adult. Reference Sparling, Howard, Dunstan and Owen74 Further studies are needed to examine the relationship between light exercise and adulthood health in growth-restricted populations. The type of physical activity (domestic/occupational versus leisure time) was also not differentiated in our analyses, which could elucidate important differences underlying the type of physical activity participants engaged in. Finally, a greater number of participants in both LBW and NBW groups, could allow for greater classification into subcategories of birthweight (very low birthweight, low birthweight, normal birthweight, high birthweight, etc) to further clarify potential differences that may be masked by the broader categorization described in this manuscript.
Acknowledgements
The authors would like to thank Catherine Gammon for her technological support, as well as the Department of Kinesiology for their support with this project.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interests
No conflicts of interest, financial or otherwise, are declared by the authors.
Ethical standards
This study was approved by the Michigan State University Institutional Review Board (STUDY00005409). All data was de-identified by BioLINCC prior to analysis.
Open access
