Characteristics of Framingham Offspring Participants With Long-lived Parents FREE

Prior research suggests that longevity runs in families. Both parents and siblings of centenarians have been shown to have significantly longer life expectancies than the average for their birth cohorts.^1- 2 More recently, it has been demonstrated that centenarian offspring have a survival advantage compared with age-matched referents whose parents died at an average life expectancy.^3- 4 Furthermore, centenarian offspring demonstrate reduced prevalence,^3- 4 as well as a delay in age at onset, for heart disease, hypertension, and diabetes,⁵ supporting the idea that the avoidance or delay of cardiovascular disease (CVD) and related risk factors facilitates survival to exceptional old age.

The multigenerational Framingham Heart Study (FHS) provides an opportunity to assess the relations between parental longevity and offspring prevalence of CVD risk factors in the community. We hypothesize that offspring with long-lived parents have more favorable risk factor profiles compared with those whose parents have average or diminished longevity. Unlike studies with cohorts enriched for longevity, the FHS is a community-based cohort study with a broad range of survival; as such, we used a cutoff of 85 years or older to define parents who are long-lived, based on the commonly used definition of oldest old.⁶

The FHS is a longitudinal study of risk factors for cardiovascular and other chronic diseases in a community-based sample of residents initially residing in Framingham, Mass. The original cohort included 5209 persons aged 28 to 62 years who have been followed up with biennial examinations from 1948 to present.⁷ The offspring cohort included 5124 persons aged 5 to 70 years who have been followed up with examinations every 4 to 8 years from 1971 to present.⁸ The FHS examination protocols are reviewed by the institutional review board of the Boston University Medical Center, Boston, Mass, and all participants signed informed consent.

The FHS offspring members were eligible for the present study (n = 2532) if both of their parents (1) participated in the original cohort of the FHS and (2) survived to 85 years or older or died prior to January 1, 2005. Offspring who had not yet attained age 30 years (n = 835) by the date of the FHS offspring examination cycle 1 (1971-1975) were excluded from the analysis because the estimation of Framingham Risk Score (FRS) requires an age of at least 30 years.⁹ Our final study sample included 1697 participants (868 women).

Variables of interest examined in the offspring cohort included age, sex, education, current cigarette smoking, systolic and diastolic blood pressures, serum total–high-density lipoprotein cholesterol (HDL-C) ratio, body mass index (BMI) (calculated as weight in kilograms divided by height in meters squared), and FRS. Risk factor data collected during FHS offspring examination cycle 1 were used in the cross-sectional analysis. Attendees of examination cycle 3 (1983-1987) provided risk factor data for the longitudinal tracking of risk factor levels between examination cycles 1 and 3. At both FHS examinations, participants completed physician-administered medical history and physical examinations, standardized questionnaires, a blood test panel, and anthropometric measurements. The number of years of education (ascertained at examination cycle 2) and current cigarette smoking (within the year prior to examination) were defined by self-report. Hypertension was defined as having a blood pressure of 140/90 mm Hg or higher or currently taking antihypertensive medication.¹⁰

We were unable to examine glucose intolerance owing to uncertain fasting status at examination cycle 1. In addition, we lacked power to examine diabetes and its progression because only 1% of the sample had diabetes at examination cycle 1. The prevalence of offspring CVD, as defined by a history of angina, coronary insufficiency, myocardial infarction, heart failure, intermittent claudication, stroke, or transient ischemic attack at baseline was very low: 2% of men in all 3 categories, 1% in women with one or no surviving parent, and 0% in women with two long-lived parents. Hence, we were not adequately powered to examine the differences in CVD prevalence between the groups.

The associations between parental longevity and CVD risk factor levels (age, sex, years of education, current cigarette smoking, systolic and diastolic blood pressures, hypertension, serum total/HDL-C ratio, BMI, and FRS) in offspring in their mid- to young adulthood were examined after adjustment for age and sex. Offspring were grouped according to whether neither (n = 705), one (n = 804), or both parents (n = 188) survived to age 85 years or older. Continuous variables were evaluated using multiple linear regression methods; least squares means were obtained, and linear trend tests across parental longevity categories were performed. For dichotomous traits, we evaluated trends in the sex- and age-adjusted proportions in each parental longevity category using logistic regression analysis. Cochran-Mantel-Haenszel tests were performed to assess the association of parental survival and risk factor categories. To examine whether the inclusion of siblings in our sample created a bias, we performed a secondary analysis on a subset of the sample that included 1 randomly selected offspring participant from each sibling group.

To compare parental longevity groups with respect to longitudinal changes in risk factor levels over a 12-year period, variables were categorized according to conventional cutpoints. These included the Joint National Commission VI criteria for optimal, normal, high-normal, and hypertensive blood pressures¹⁰ and the World Health Organization BMI criteria for normal weight, overweight, and obesity. Smoking cessation was examined in current cigarette smokers at baseline. Serum total/HDL-C ratio was categorized using sex-specific baseline tertiles (<4.2, 4.2-5.4, and ≥5.5 in men and <3.0, 3.0-3.8, and ≥3.9 in women).¹¹ The FRS is a sex-specific estimator of 10-year coronary heart disease risk based on age, total and HDL-C levels, systolic and diastolic blood pressures, diabetes, and current cigarette smoking. The FRSs were categorized according to sex-specific baseline tertiles of FRS (≤1, 2-3, and ≥4 in men and ≤−1, 0-1, and ≥2 in women). For the present analyses, FRSs did not account for age so that the same tertile cutpoints could be used in the definition of progression. For each variable, an indicator of progression from the baseline risk category to a higher follow-up risk category was constructed for use as the dependent variable in multivariable-adjusted logistic regression analysis.

With the exception of FRS, participants falling into the highest risk category at baseline were considered not to be at risk for progression and were excluded from the analysis for that particular risk factor. In the case of the FRS, all individuals free of CVD at baseline were considered to be at risk of progression, defined as (1) having experienced a confirmed coronary heart disease event during the follow-up period or (2) an increase in FRS category from baseline to the follow-up examination. Individuals at baseline taking antihypertensive or lipid-lowering medications were considered to have hypertension or hyperlipidemia and were excluded from the blood pressure or the cholesterol progression analysis, respectively. Logistic regression analysis was used to examine the association between parental longevity category and risk factor progression between examination cycles 1 and 3. Models were adjusted for sex and baseline age and risk category. Parental longevity was modeled using separate indicator variables for having one or both parents survive to 85 years or older, each of which were compared with having neither parent survive to 85 years or older. Parental longevity was also modeled as one 3-level variable to test for trend across parental longevity groups. All statistical analyses were performed using SAS version 8.02 (SAS Institute Inc, Cary, NC). We considered P≤.05 (2-sided) to be statistically significant.

In our study sample of 1697 offspring, the proportion with neither, one, or both parents surviving to 85 years or older was 42%, 47%, and 11%, respectively. The overall mean age at examination cycle 1 was 40 years; across categories of parental survival, each additional surviving parent was associated with about a one-half year younger mean offspring age (Table 1) (P = .02).

Table 1 gives the sex- and age-adjusted examination 1 (baseline) characteristics of offspring participants based on their number of parents surviving to 85 years or older. We observed that baseline offspring risk factor levels were significantly lower as the number of parents surviving to 85 years or older increased. The offspring of the longest-lived parents were more likely to be women (55%, adjusting for age) and to have completed more years of education. There was a statistically significant trend toward more favorable offspring risk factor characteristics across parental survival groups in the mean systolic and diastolic blood pressures, hypertension prevalence, mean serum total/HDL-C ratio, and mean FRS. No significant difference in BMI between offspring in the 3 parental survival groups was noted. We found a statistically significant interaction between sex and parental survival group for smoking; in men, current smoking at baseline was associated with parental survival, but in women, no difference was seen across parental survival categories. No other interactions with either sex or age were found; therefore, only smoking is presented separately in Table 1 for men and women.

In the group with only one parent surviving to 85 years or older, no significant differences in risk factor levels in offspring of long-lived mothers vs long-lived fathers were observed. For instance, the offsprings' mean systolic blood pressures were less than 1 mm Hg different regardless of which parent survived (Table 2).

The analysis given in Table 1 was repeated on a subset of the sample including only 1 random member from each sibling group. In the full sample, more than 50% of participants had no sibling enrolled in the offspring study, less than 30% had 1 sibling, approximately 10% had 2 siblings, and the remainder had 3 or more enrolled siblings. In analyses restricted to only 1 sibling, we observed the same trends for lower risk factor levels with increasing number of long-lived parents (data not shown). For systolic blood pressure and current cigarette smoking, the trend was more pronounced in the subset with siblings removed than it was in the primary analysis.

Baseline risk categories by parental survival group are given in Table 3. Variables were categorized according to conventional cutpoints as described in the “Methods” section.

The percentage of those individuals with optimal or normal blood pressure, total/HDL-C ratio, and low FRS was highest in those with both parents surviving to 85 years or older. The relations for BMI were less clear; however, fewer obese individuals had both parents survive.

We assessed the longitudinal progression of risk factors to higher risk categories according to parental longevity from examination cycles 1 to 3, a period spanning approximately 12 years. Only 5% of the baseline sample had died in the interim (6% of the group with neither parent surviving and 4% of the groups with either or both parents surviving to 85 years or older). Of those alive at the time of examination cycle 3, 1319 members (82%) of the baseline sample attended the third examination.

Risk factor category progression is shown in Table 4. Individuals with both parents surviving to 85 years or older were less likely to have progression of blood pressure or FRS, with statistically significant trends across the 3 parental survival groups. No difference was seen in the likelihood of smoking cessation across parental longevity groups. There was a suggestion of a trend toward greater progression in BMI in the offspring of longer-lived parents (P = .07) that became significant in a secondary analysis restricted to nonsmokers at baseline (odds ratio, 1.36; P = .03).

A “dose-response” relation of parental longevity with progression of the FRS is demonstrated in the Figure. The odds of progression were highest for offspring with neither parent surviving and lowest for offspring with both parents surviving to 85 years or older (P value for trend, <.001). We did not observe significant effect modification by sex and baseline age in the odds for offspring risk factor progression by parental longevity status. A secondary analysis, with additional adjustment for years of education, revealed that education was not a confounder, and the magnitude of the associations was unchanged (data not shown).

Compared with offspring whose parents died at earlier ages, the offspring of long-lived parents of the FHS original cohort had significantly lower levels of cardiovascular risk factors in middle age, including a lower proportion of current smokers and individuals with hypertension, lower mean systolic and diastolic blood pressures, and lower serum total/HDL-C ratio. They also had markedly lower FRSs, a summary score based on the joint contribution of traditional cardiovascular risk factors. In addition, we demonstrated that they were less likely to have progression of blood pressure and FRS over 12 years of follow-up. Our findings suggest that individuals with long-lived parents have more advantageous cardiovascular risk profiles in middle-age compared with those whose parents died younger and that the risk factor advantage persists over time.

To our knowledge, our study is the first to examine cardiovascular risk factors in the offspring of longer-lived individuals using independent and validated measurements of cardiovascular risk factors. Other large cohort studies, such as the PRIME study (Prospective Epidemiological Study of Myocardial Infarction),¹² the National Heart, Lung, and Blood Institute's twin cohort study,¹³ and the Primary Prevention Study,¹⁴ have relied on offspring self-report of family history and age at parental death. Nevertheless, our findings of an inverse relation between systolic and diastolic blood pressures and parental survival are consistent with the aforementioned studies.

In addition, our study expands the understanding of the familiality of longevity by examining the progression of CVD risk factors and FRS over time. We demonstrated a decline in the rate of the progression of blood pressure comparing offspring with neither, one, or both parents surviving to 85 years or older. Furthermore, our study expanded prior research into the familiality of longevity by demonstrating the dose-dependent relation between parental longevity and FRS in offspring.

The overall prevalence of risk factors in our study is low compared with other longevity studies.^3- 4 In part, this is due to the parental age cutoff of 85 years; the offspring are relatively young (mean age, 40 years) and may not have developed risk factors as of yet.

It is interesting to note the trend toward younger offspring ages among those with both parents surviving to 85 years or older. A prior study demonstrated that centenarian women were much more likely to have children in their 40s compared with women who survived to age 73 years.¹⁵ It is hypothesized that the ability to have children later in life may be a proxy for delayed menopause and, perhaps, delayed aging. Another possibility is that delayed childbirth may be a surrogate for socioeconomic status.¹⁶

The offspring of long-lived parents of the FHS original cohort achieved significantly higher education levels than those whose parents died at younger ages. Education has been used as a proxy for socioeconomic status, with higher levels being associated with lower all-cause mortality.¹⁷ Educational attainment is often familial, so higher levels of offspring education may reflect higher levels of parental education and, in turn, higher socioeconomic status.

Male offspring with long-lived parents were less likely to have smoked. Smoking has been shown to be familial,^18- 19 so it may be that the shorter-lived parents, like their offspring, were more likely to have smoked, contributing to premature mortality. Interestingly, we did not demonstrate the inverse association between long-lived parents and smoking status in female offspring. Prior research has shown sex-related differences in smoking. Compared with men, women, on average, start smoking at younger ages but also are more likely to cut back or quit as they get older.²⁰ Sex-related differences in smoking behavior may partially explain why the association between not smoking and long-lived parents was demonstrated only in men.

Our results did not demonstrate a difference in BMI among the 3 parental groups if we examined BMI as a cross-sectional continuous variable. The association between BMI progression (using categories of normal weight, overweight, and obese) and long-lived parents, while suggestive of a trend, was not statistically significant (P = .07). In a secondary analysis restricted to baseline nonsmokers, the direct association between parental survival and BMI progression achieved statistical significance (P = .03). The apparent “protective” association of weight progression with parental survival is counterintuitive, given reports that increased BMI is associated with CVD and cancer.^21- 23 Alternatively, it has been demonstrated that the association between BMI and longevity is J-shaped^24- 25; very low BMI levels also are associated with higher health risks. It may be that the lack of progression in the offspring of the less long-lived parents may represent weight loss secondary to disease. In addition, the mean BMI of the participants at offspring examination cycle 1 (1971-1974) was 26. It is uncertain if our BMI results would be generalizable to a cohort with a higher prevalence of overweight and obese individuals.

Prior studies have shown contradictory findings on whether offspring survival was associated with maternal^13,26 vs paternal longevity.¹⁴ In our study, we did not observe significant differences attributable to the sex of the long-lived parent.

One of the major strengths of our study is the use of data from the FHS cohort, which has been well characterized over 5 decades. Furthermore, since the data were collected and recorded longitudinally on parental and offspring generations, they are less susceptible to recall or survivor bias.

The study cohort was white and predominantly middle class; thus, the generalizability of our findings to other races/ethnicities and economically disadvantaged populations remains to be determined. However, in contrast to some of the other studies of longevity, this study was community based. In addition, the requirement of having two parents enrolled in the FHS may introduce a healthy volunteer bias, thus reducing the generalizability of our findings to the overall population. In addition, we may have lacked power to demonstrate differences by parental survival category for cross-sectional BMI analyses and longitudinal analyses of smoking cessation in current smokers at baseline. Similarly, our study was not sufficiently powered to detect small to modest maternal vs paternal variation in offspring characteristics by parental survival category.

In conclusion, the offspring of longer-lived parents have a lower prevalence of cardiac risk factors in middle age. In addition, they are less likely to develop progression of blood pressure and FRS category over time. There are well-established genetic contributions to each of the risk factors that we have examined that may partially explain the reduced risk factors for those with long-lived parents. Better understanding of genetic variation in cardiovascular risk factors and longevity eventually may be helpful for disease prevention and treatment strategies in the community.

Correspondence: Dellara F. Terry, MD, MPH, Geriatrics Section, Boston Medical Center, 88 E Newton St, Robinson 2, Boston, MA 02118 (laterry@bu.edu).

Accepted for Publication: October 11, 2006.

Author Contributions: Dr Evans had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Terry, Evans, Pencina, Murabito, Vasan, and Benjamin. Acquisition of data: Kelly-Hayes, Levy, and D’Agostino. Analysis and interpretation of data: Terry, Evans, Pencina, Murabito, Vasan, Wolf, D’Agostino, and Benjamin. Drafting of the manuscript: Terry, Evans, Murabito, Vasan, and Benjamin. Critical revision of the manuscript for important intellectual content: Terry, Evans, Pencina, Murabito, Vasan, Wolf, Kelly-Hayes, Levy, D’Agostino, and Benjamin. Statistical analysis: Evans, Pencina, and D’Agostino. Obtained funding: Wolf, and D’Agostino. Administrative, technical, and material support: Wolf and Kelly-Hayes. Study supervision: Benjamin.

Financial Disclosure: None reported.

Funding/Support: This work was supported by contract NO1-HC-25195 from the National Heart, Lung, and Blood Institute, grant 5R01-NS-17950 from the National Institute of Neurological Disorders and Stroke (Dr Wolf), grant 1K08-AG022785 and Paul B. Beeson Career Development Award 1K23-AG026754 from the National Institute on Aging (Dr Terry), grant K24-HL-04334 from the National Heart, Lung, and Blood Institute (Dr Vasan), and R01-AG028321 from the National Institute on Aging (Dr Benjamin), National Institutes of Health.

Previous Presentation: This study was presented as a poster at the American Geriatrics Society Annual Meeting; May 4, 2006; Chicago, Ill.