|Home | About | Journals | Submit | Contact Us | Français|
Data relating parental history of stroke to stroke risk in offspring remain surprisingly inconsistent, largely due to heterogeneity of study design, and the absence of verified, as opposed to historical, data on parental stroke status.
We determined if prospectively verified parental occurrence of stroke increased incident stroke risk among offspring in a community-based sample by studying 3443 stroke-free Framingham Offspring (53% female, mean age 48±14 years) with verified parental stroke status (by age 65 years), who attended the 1st, 3rd, 5th and/or 7th Offspring examinations, and were followed for up to 8 years after each baseline examination. Over up to 11,029 such person-observation periods (77,534 person-years), we documented 106 parental strokes by age 65, and 128 offspring strokes (74 parental and 106 offspring strokes were ischemic). Using multivariable Cox models, adjusted for age-, sex-, sib-ship and baseline stroke risk factors, we observed that parental stroke, both all-stroke generally, and ischemic stroke specifically, was associated with an increased risk of incident stroke of the same type in the offspring (HR 2.79, 95% CI: 1.68–4.66; p<0.001 for all stroke, and HR 3.15, 95% CI: 1.69–5.88; p<0.001 for ischemic stroke). This was true for both maternal and paternal stroke.
Documented parental stroke by age 65 years was associated with a three-fold increase in risk of offspring stroke. This increased risk persisted after adjustment for conventional stroke risk factors. Thus, verified parental stroke may serve as a clinically useful risk marker of an individual’s propensity to stroke.
Stroke remains a leading cause of morbidity and mortality. Several environmental risk factors for stroke have been identified, but there is also a substantial heritable component to stroke risk.1, 2 Several known monogenic disorders increase stroke risk and account for a familial aggregation within certain families; typically these individuals develop stroke at a young age.3 However, whereas parental coronary heart disease (CHD) occurring at <60 years is accepted as a marker of increased CHD risk in the offspring,4, 5 it has proved harder to definitively establish a genetic component for stroke risk in the general population with surprising inconsistencies in published results.6–14 Results have ranged from a doubling of risk in the offspring11 to no observed impact of parental history of stroke15, 16 to an effect limited to certain age-, sex- groups,6 an effect restricted to paternal stroke12, 14 or an effect confined to maternal stroke.13
One reason for this variability in findings may be misclassification of parental stroke status. Prior studies examining the familial aggregation of stroke have either used family history questionnaires administered to a proband,8, 17–19 direct questioning of pedigree members20 or data from death certificates.21 Each of these methods has significant limitations: family history questionnaires underestimate stroke occurrence in family members, direct questioning of probands would favor ascertainment of non-fatal over fatal stroke whereas death certificate data underestimate the risk of non-fatal stroke. All three methods result in incorrect inclusion of some non-stroke events.22 The Framingham Study has the unique resource of directly verified data on stroke occurrence in both parents and offspring who belong to the Framingham Original and Offspring cohorts respectively. The only previous investigation examining the verified occurrence of parental stroke (as opposed to a history of parental stroke) to the risk of offspring stroke was an earlier report from the Framingham Study.23 These analyses suggested the value of verified occurrence of stroke over a family history. There was no association between reported parental stroke death and the combined end-point of incident stroke or transient ischemic attack (TIA) in Original cohort participants but there was a trend towards an association between verified parental stroke/TIA in the Original cohort and the risk of stroke/TIA in the offspring. However, the small number of stroke events available for analysis at the time of that initial report limited the power of that study, and the results failed to reach statistical significance.23 We now reexamine and expand these prior analyses using the additional cases accrued since our previous report.
Familial aggregation may also vary by stroke subtype and studies that examined the heritability of all-stroke (without classifying by stroke subtypes) may therefore fail to detect a real familial aggregation.24 The subtype of ischemic stroke (IS) is of greatest interest since this constitutes the largest proportion of total stroke burden in the population. Finally, risk may vary based on the age at which the parental stroke occurred; some (but not all) studies have observed an increased risk only among individuals below age 65 or 70 years.8, 19, 24–27 In the current investigation we prospectively assessed the relationship between parental occurrence of stroke by age 65 years and risk of incident stroke in the offspring. We examined the risks associated with the specific subtypes of IS and atherothrombotic brain infarction (ABI) as these are the most common subtypes. Finally, we adjusted for the impact of traditional stroke risk factors which had been systematically assessed at each baseline evaluation of the Framingham Original and Offspring cohorts.
The Framingham Heart Study is an ongoing prospective study of cardiovascular disease and its risk factors in an Original (“parental”) cohort of 5209 women and men who were aged 28 to 62 years at the time of enrollment in 1948. In 1971, offspring of the original cohort and the spouses of these offspring (n = 5124, age 12–58 years, 3548 biological offspring, 1576 offspring spouses) were enrolled in the Framingham Offspring Cohort. These participants have been followed every 4 years with the exception of the second examination cycle which occurred 8 years after the first. The design of the Framingham Heart Study (Original and Offspring Cohort) and methods of risk factor measurements have been described earlier.28–30 In the present investigation, Framingham Offspring Cohort participants free of prevalent stroke, who attended the 1st, 3rd, 5th or 7th offspring examinations were followed for up to 8 years or until the start of the next follow-up period; each such person-interval will be called an person-observation period, or simply an ‘observation period.’ We used 8-year observation periods because parental stroke status and presence or absence of conventional stroke risk factors could vary between observation periods. However, the occurrence of stroke was ascertained in an ongoing, prospective manner and not only once every 8 years.31 Offspring cohort participants who had one or more parents enrolled in the Framingham Original cohort were included in our study if the enrolled parent had either sustained a prospectively identified and verified clinical stroke by age 65 or was known to be stroke-free at this age. We chose this age cut-off to define parental stroke status as it had been proposed in previous studies.8 The number of participants who were alive at the start of each baseline examination, the number attending each examination and the numbers excluded (for absence of parental information, for prevalent stroke at baseline, and for lack of follow-up information on occurrence of incident stroke) are shown in Table 1. Since the Offspring cohort includes not only the biological children of the Original cohort participants but also their spouses (whose parents may not have been enrolled in the Framingham Original cohort) there is a substantial difference between the numbers who attended each baseline examination and those included in our study. We included a total of 3,443 Offspring who provided information over 11,029 observation periods (77,534 person-years). Risks were determined separately for persons in whom parental stroke status by age 65 years was known in both parents (n=6633 observation periods, 46517 person-years), in fathers only (n=8069 observation periods,) and in mothers only (n= 8473 observation periods). The parental information on these 3,443 Offspring was drawn from 2,630 Original cohort participants. This study was approved by the Institutional Review Board of Boston Medical Center and all subjects gave informed consent.
Stroke was defined as an acute onset focal neurological deficit of presumed vascular etiology, persisting for ≥ 24 hours. All events were adjudicated by a panel of 2 neurologists. Details of stroke surveillance and protocol for determining the final diagnosis, type, localization and severity of stroke have been published; TOAST criteria32 have been used since 1993 and events that occurred prior to 1993 have been reclassified using TOAST criteria wherever feasible.33, 34 A transient ischemic attacks (TIA) was defined as an episode of rapid onset focal neurological dysfunction attributed to focal cerebral ischemia, with resolution within 24 hours. Individual stroke subtypes were categorized based on standard diagnostic criteria and considered clinical features, imaging studies and other laboratory data such as noninvasive vascular studies, cardiac assessment for possible embolic sources and, when available, autopsy data. Ischemic stroke (IS) was diagnosed if a focal neurological deficit was documented and either the imaging showed no hemorrhage, imaging showed an ischemic infarct that correlated with the clinical deficit, or an ischemic infarct was documented at autopsy. ISs were further classified as atherothrombotic brain infarcts (ABI) if no cardiac source of embolus was found; this category included large-artery infarcts, lacunar infarcts and infarcts of unknown origin.
We adjusted our analysis for age-, sex-, sib-ship among the Offspring participants and for the covariates previously described as a part of the Framingham Stroke Risk Profile (FSRP). The latter include age, sex, systolic blood pressure (SBP), antihypertensive therapy, diabetes, smoking status, history of cardiovascular disease (CVD) other than stroke, and the presence of atrial fibrillation or left ventricular hypertrophy on the electrocardiogram (EKG-LVH). Risk factor data for the Offspring cohort was gathered at the baseline offspring examination (1st, 3rd, 5th or 7th) that marked the beginning of each person-observation period. For the Original cohort we used covariate data from the examination that the parent attended when he or she was closest to age 50 years (±5 years). Current smoking was defined as cigarette smoking during the past year, presence of diabetes mellitus was defined by a recorded fasting blood glucose >126 mg/dL (7 mmol/L), a previous diagnosis of diabetes mellitus, or the use of an antidiabetic drug or insulin. An average of two physician- recorded systolic blood pressure measurements was used. Previous CVD events included a diagnosis of coronary heart disease, congestive heart failure, or peripheral vascular disease. The diagnosis of atrial fibrillation and EKG-LVH was based on consensus panel review of a standard 12-lead EKG obtained at the examination, as well as data from a careful history at each study examination and diligent tracking of all relevant medical records on each participant; however the diagnosis may have been missed in some persons with paroxysmal atrial fibrillation.
Participants were followed until they developed the outcome of interest (stroke or TIA), died or were lost to follow-up. Cox proportional hazards models were used to determine the risk of incident offspring stroke over each observation period. Increased risk conferred by early parental occurrence of stroke was estimated after adjusting for age-, sex-, and sib-ship among offspring. We used generalized estimating equations to adjust for sib-ship and for the correlated nature of an individual’s risk over several successive observation periods (Model A). In secondary analyses we adjusted for baseline stroke risk factors in the offspring (Model B) and examined the association of parental stroke with early onset stroke in the offspring occurring before the age of 65 years (Model C). We conducted analyses separately for maternal and paternal stroke, by stroke subtype in parent and offspring and including TIA as well as completed strokes. We also studied the combined outcome of ABI and TIA since all ABI and most TIA share a common pathophysiological substrate of either large artery thrombosis or small artery occlusion, differing only in the duration of clinical symptoms. For the analyses of risk associated with parental (including both maternal and paternal) stroke, we required that stroke status information be available for both parents; for the analyses of the separate risks conferred by maternal and paternal strokes we required only that information for that specific parent be available. For analyses assessing familial risk within stroke subtypes, participants who developed an alternative stroke subtype continued to be followed for the development of IS. Finally, we examined the risks associated with parental stroke status within quintiles of baseline FSRP profile score in the offspring (Model D). In tertiary exploratory analyses we also examined these associations after additional adjustment for midlife (age 50±5 years) parental stroke risk factors (Model E) and within sex-specific offspring subgroups (Model F).
We directly documented 128 strokes (106 of which were ischemic and 81 of which were ABI) and 29 initial incident TIAs among the 3,443 Framingham Offspring participants in our study sample over a total of 11,029 observation periods. We had also prospectively identified or verified an incident clinical stroke by age 65 in 106 parents (74 of these strokes were ischemic, 55 ABI) and an initial incident TIA by age 65 years in 22 parents. Of these 50 strokes (35 ischemic, 30 ABI) and 2 isolated TIA were in mothers and 56 strokes (39 ischemic, 25 ABI) and 20 TIA were in fathers.
The distribution of baseline demographic and stroke risk factor characteristics in Offspring with and without parental stroke by age 65 is summarized in Table 2. In comparison to offspring without parental stroke, offspring with verified parental stroke were younger, more likely to be men, and had higher mean SBP, greater use of antihypertensive medication, and a higher prevalence of diabetes, smoking and electrocardiographic LVH. Aggregate FSRP scores were also higher in the group with parental stroke. The prevalence of atrial fibrillation and prior CVD did not differ between the two groups. Table 2b describes the midlife (age 50±5 yrs) distribution of stroke risk factors in the parents.
Table 3a describes the hazard ratios for incident stroke by age 65 years among offspring participants with parental stroke as compared to offspring whose parents were documented to remain stroke-free till age 65 years (Model A). The results are presented by subtype of parental and offspring stroke. Table 3b shows results after adjusting for baseline stroke risk factors in the Offspring (Model B) and Table 3c presents the risk of early Offspring stroke occurring by age 65 years (Model C). We observed that offspring with documented parental history of any stroke type by age 65 years had a two to three fold increased risk of stroke and this risk was greatest for the subtype of ABI, although the risk was also increased for all-stroke, IS and the combined ABI and TIA categories of offspring stroke.
Adjustment for baseline levels of conventional stroke risk factors attenuated but did not abolish these associations. To assess if the observed familial aggregation was restricted to Offspring with a lower risk-factor burden we compared the cumulative incidence of stroke by age 65 years within five subgroups of Offspring participants grouped by quintile of baseline FSRP score at the 1st, 3rd, 5th or 7th Offspring examination (Model D). The results are shown in Figure 1. The higher risk in persons with parental stroke was seen across all quintiles of baseline FSRP score with the greatest impact observed among persons in the highest quintile of stroke risk. Additional adjustment for midlife stroke risk factors in the parent (at age 50 years) also did not abolish the observed association of parental and offspring stroke (Model E, Supplementary Online Appendix: Table A).
The relative risk (hazard ratio) associated with parental stroke was highest when we restricted our analyses to early stroke (occurring before age 65 yrs) in the Offspring. Thus, in analyses adjusted for baseline stroke risk factors, parental stroke by age 65 approximately doubled the risk of a stroke in the offspring (HR 2.21; 95% CI: 1.32–3.70) but increased the risk of an offspring stroke before age 65 years nearly 4-fold (HR 3.79; 95% CI 1.90–7.58).
Separate analyses were performed for paternal and maternal history of stroke; these results are presented in Tables 3a–c. Offspring with a positive paternal history of any stroke type had a 2 to 3 times greater risk of ever developing any stroke, IS, ABI, or a combined outcome of ABI and TIA and a 4 times greater risk of developing a stroke by age 65 years. An increased risk was also associated with paternal IS although the impact of paternal ABI or ABI and TIA on offspring stroke risk was less consistent, likely due to smaller numbers in the subgroup analyses. Maternal history of any stroke was again associated with a 2 to 3 fold increase in the risk of offspring strokes. This was true for any offspring stroke, for IS, ABI, or a combined outcome of ABI and TIA but again the results were less consistent when we considered individual subtypes of maternal stroke. As observed for all parental strokes, the effects of paternal and maternal strokes on offspring stroke persisted after adjustment for baseline stroke risk factors in the offspring and were strongest when considering offspring risk of early-onset stroke. In sex-specific analyses (Model F) paternal stroke was related to the risk of offspring stroke in both men and women, whereas maternal stroke was significantly associated with the risk of offspring stroke in women but not in men (Supplementary Online Appendix: Table B).
In the two-generational, community-based, Framingham Study cohort followed prospectively for incident stroke and stroke risk factors over the past 4 decades, we observed that prospectively documented parental occurrence of stroke by age 65 years, was associated with a substantially increased risk of incident stroke in the offspring. This association was evident in analyses of all-stroke as well as in analyses limited to individual stroke subtypes, being consistent across sub-groups categorized by subtypes of parental or offspring stroke; subtypes assessed include IS, ABI, and ABI and TIA. The impact of parental stroke was greatest on the offspring risk of an early stroke occurring before age 65 years, a finding consistent with previous studies.8
The detection of a familial aggregation in stroke risk leaves open the question of the relative contribution of shared genetic and environmental factors. Previous case-control and cohort studies examining the familial aggregation of stroke have been largely based on twin studies and sib pair analyses; there is a greater probability of confounding by shared environmental exposure in these studies. Observed associations of parental and offspring stroke have also been partly attributed to familial clustering of stroke risk factors such as hypertension and left ventricular hypertrophy.25, 26, 35 In case-control studies it is difficult to accurately determine the levels of stroke risk factors present prior to the event since measurements such as blood pressure may be altered by physiological consequences or pharmacological interventions following a stroke. The current analyses utilized the availability of stroke risk factor data that had been prospectively ascertained prior to the event to adjust for baseline stroke risk factors in the offspring and showed the risk associated with parental stroke could not be entirely attributed to the inheritance and expression of baseline stroke risk factors such as hypertension in the offspring with parental stroke since the effect persisted after adjustment for stroke risk factor levels in the offspring. Moreover, the increased risk conferred by parental stroke status was observed at each quintile of offspring 10-year stroke risk as determined by the FSRP score.
Our findings are consistent with prior studies reporting a marginally stronger association of paternal stroke (relative to maternal stroke) with offspring stroke.12, 14 One prior report had described a stronger association of maternal (compared to paternal) stroke with offspring stroke,6 an observation that our study failed to confirm. One difference between the two studies is that Touze et al., questioned the proband to ascertain the occurrence of parental strokes whereas stroke events were directly verified in the Framingham parental generation. It is possible that mothers may share their medical history, particularly with regard to transient events such as TIA, more readily with their offspring than would fathers. In a previous report from the Framingham study examining the accuracy of offspring reports of parental cardiovascular disease, an offspring’s report of maternal stroke had a greater sensitivity and positive predictive value than an offspring’s report of paternal stroke.22 Touze et al., used a cross-sectional study design that may have detected the greater probability that women, with their higher life-expectancy, would express an inherited propensity to stroke.6 Our study design which used a survival analysis is less likely to be affected by the difference in survival between women and men. Finally, Touze et al., restricted their analysis to probands in whom parental stroke status was available for both mothers and fathers,6 but repeating our analyses to exclude Offspring with only one parent enrolled in the Framingham Original cohort did not alter our results (data not presented). We did observe that the association between maternal and offspring stroke was stronger in female offspring compared to male offspring, which was in agreement with the results of a meta-analysis by the same authors.7
The strengths of our study are the community-based sample, the prospective verification of both parental and offspring stroke status, and the careful classification of stroke subtypes based on neurological examination, record review and imaging. Further, the impact of vascular risk factors could be accurately assessed since these had been systematically measured prior to the stroke, both in the parents and in the offspring. Limitations of this study include the overwhelmingly European origin of the study sample and the limited number of events among individual stroke subtypes which restricted our ability to conduct separate subgroup analysis for stroke subtypes of cardioembolic and lacunar strokes although the latter was included in the category of atherothrombotic strokes. Finally we did not adjust for newer stroke risk factors such as plasma homocysteine and CRP levels. It remains possible that the observed aggregation may be attributable to familial aggregation of unidentified environmental, behavioral, and lifestyle- related factors.
Of particular preventive interest, although family history or familial occurrence of stroke is appropriately considered to be a non-modifiable risk factor for stroke, we find that the level of conventional risk factors amplifies the increased stroke risk associated with a positive family history, particularly in the upper two quintiles of the FSRP (Figure 1). Thus, attention to control of modifiable risk factors for stroke prevention, particularly in the presence of a positive familial occurrence (or history) of premature stroke, is strongly supported by these data.
Nevertheless, the demonstrated strong association of parental and offspring strokes and the particularly strong association demonstrable between IS in parent and offspring suggest that there is indeed a substantial genetic contribution to the risk of all-stroke and IS. This is unlikely to be due to shared environmental factors alone since the increased risk persisted after adjustment for levels of conventional stroke risk factors. Candidate gene studies36 and more recently genome-wide association (GWA) analyses37, 38 have identified several putative genes that may underlie this genetic propensity. It is probable that there are genes increasing the risk of all stroke types and other genes that specifically increase the risk of IS and IS subtypes. The Framingham study is currently relating 550K GWA data to the risk of incident stroke; the observed strong familial aggregation suggests that such efforts should prove fruitful. Finally, the three-fold, independent, increase in risk associated with premature parental stroke (by age 65 years in our study) suggests that this simple measure might prove a useful risk marker in clinical risk prediction. ‘Parental stroke status’ and ‘age of parent at stroke’ may be measures worth incorporating in future revisions of the Framingham and other stroke risk prediction algorithms.
The Framingham Heart Study has prospectively verified data on occurrence of stroke across two generations of participants, the Original (parental) and Offspring cohorts. We studied incident stroke risk among 3443 stroke-free Offspring (53% female, mean age 48±14 years) with verified parental stroke status (by age 65 years). Using multivariable Cox models, adjusted for age-, sex-, sib-ship and baseline stroke risk factors, we observed that over a period of 77,534 person-years, verified parental stroke by age 65 resulted in a near 3-fold, independent, increase in risk of offspring stroke (HR 2.79, 95% CI: 1.68–4.66; p<0.001 for all stroke, and HR 3.15, 95% CI: 1.69–5.88; p<0.001 for ischemic stroke). This was true for both maternal and paternal stroke. The increased risk persisted after adjustment for conventional stroke risk factors. Thus, parental stroke status might serve as a simple, clinically useful, aggregate measure of an individual’s hereditary propensity to stroke. It has been suggested that for many polygenic diseases and traits, testing for multiple risk alleles may not improve upon the use of ‘family history’ as a risk marker.
Funding Sources: This work was supported by the Framingham Heart Study’s National Heart, Lung, and Blood Institute contract (N01-HC-25195) and by grants from the National Institute of Neurological Disorders and Stroke (R01 NS17950) and from the National Institute on Aging (R01 AG16495; AG08122; AG033193 and AG031287). Dr. Debette is supported by a Fulbright grant and received an award from the Bettencourt-Scheuller Foundation.
Conflict of Interest Disclosures: None