Genetic Liability in Stroke
A Long-Term Follow-Up Study of Danish Twins
Background and Purpose— Few studies have assessed the overall importance of genetic factors on stroke risk, and the results have been contradictory. We used a large, population-based twin register and nationwide registries of death and hospitalization with long-term follow-up to estimate the effect of genetic factors on the risk of stroke.
Methods— Through the population-based Danish Twin Register, we identified same-sex twin pairs born in 1870 through 1952 for whom at least 1 twin was recorded under a stroke diagnosis in the Register of Causes of Death or the Danish National Discharge Register. From the day of the first stroke event in each twin pair, the live co-twins were followed up for stroke. In survival analyses, we estimated the age- and sex-adjusted effect of zygosity on the risk of stroke death or hospitalization for stroke. Concordance rates, tetrachoric correlations, and heritability were also assessed.
Results— Thirty-five of 351 monozygotic pairs (10%) and 34 of 639 dizygotic pairs (5%) were concordant for stroke death. The age- and sex-adjusted relative risk of stroke death in monozygotic compared with dizygotic co-twins was 2.1 (95% CI, 1.3 to 3.3). The probandwise concordance rates were 0.18 (95% CI, 0.14 to 0.22) for monozygotic and 0.10 (95% CI, 0.08 to 0.13) for dizygotic pairs. Thirty-three of 309 monozygotic pairs (11%) and 39 of 560 dizygotic pairs (7%) were concordant for stroke hospitalization or stroke death. The age- and sex-adjusted relative risk of stroke hospitalization or stroke death in monozygotic compared with dizygotic co-twins was 1.5 (95% CI, 0.9 to 2.4). The probandwise concordance rates were 0.19 (95% CI, 0.15 to 0.24) for monozygotic and 0.13 (95% CI, 0.10 to 0.16) for dizygotic pairs. The heritability estimates were 0.32 for the liability to stroke death and 0.17 for the liability to stroke hospitalization or stroke death.
Conclusions— The observed increased risk of stroke death and stroke hospitalization in monozygotic compared with dizygotic co-twins suggests that genetic factors increase the risk of stroke and that the size of this effect is moderate.
Stroke is one of the most common causes of death and disability in the Western part of the world. Most stroke cases have been ascribed to lifestyle factors, but in recent years, an increasing interest in genetic risk factors for stroke has emerged. However, little is known about the overall influence of genes on the risk of stroke. The occurrence of hypertension, diabetes, and coronary heart disease, which are important risk factors for stroke, is influenced by genetic factors. Furthermore, genes related to platelet and arterial endothelial function have recently been proposed as independent stroke risk factors.1–5⇓⇓⇓⇓ Thus, genes might influence the occurrence of stroke either directly or through a genetic contribution to other stroke risk factors.
Twin studies provide an opportunity to assess the relative importance of genetic factors in a disease.6 A higher degree of similarity for a trait in monozygotic twin pairs compared with dizygotic twin pairs indicates a genetic contribution to the etiology. However, the largest twin study on stroke risk until now comprised only 7 monozygotic and 12 dizygotic pairs with both twins registered dead as a result of stroke.7 Concordance rates for stroke death were similar in monozygotic and dizygotic pairs.
The importance of hereditary factors on stroke risk has also been assessed in family studies, but the results have been contradictory, at least partly because of methodological differences between the studies, potential recall bias in assessment of family history, and inclusion of different stroke subtypes.8–14⇓⇓⇓⇓⇓⇓ Furthermore, traditional family studies cannot disentangle the effect of genetic factors and common environment.
To overcome some of these methodological problems, we used a population-based twin register and nationwide registries of death and hospitalization to estimate the relative effect of genetic factors on the risk of stroke.
Subjects and Methods
We performed a follow-up study of Danish twins born in 1870 through 1952. The twins were identified through the population-based Danish Twin Registry. All Danish citizens have a unique civil registration number recorded in the Central Person Registry. By means of these registration numbers, all twins were followed up for stroke through linkage with the Danish National Discharge Registry and the Registry of Causes of Death.
The Danish Twin Registry is nationwide and population based. The registry was established in 1954 and comprises same-sex twins born in 1870 through 1930 and surviving 6 years.15 Recently, the birth cohorts from 1931 through 1952 were included in the registry, which now holds data on >32 000 twin pairs born in 1870 through 1952 and surviving 6 years. All twins in the registry were ascertained independently of any disease. Zygosity was established through a questionnaire on the degree of similarity between twins in a pair. The validity of this zygosity classification has been evaluated by comparison with blood group determinants, and the misclassification rate has been found to be <5%.16
Information on all deaths among Danish citizens has been recorded in the Registry of Causes of Death since 1943. The register comprises date of death and causes of death according to the International Classification of Diseases (ICD). Twins with an ICD code of intracerebral hemorrhage, cerebral infarction, or unspecified stroke during follow-up were considered cases of stroke. The included ICD codes were as follows: intracerebral hemorrhage, codes 331 (ICD-6 and -7), 431 (ICD-8), and I61 (ICD-10); cerebral infarction, codes 332 (ICD-6 and -7), 432, 433, 434 (ICD-8), and I63 (ICD-10); and unspecified stroke, codes 300 (ICD-5), 334 (ICD-6 and -7), 436 (ICD-8), and I64 (ICD-10). Twins with a diagnosis of subarachnoid hemorrhage were excluded.
The Danish Twin Registry comprises 11 564 same-sex twin pairs with known zygosity and with both twins being alive on January 1, 1943. Within this cohort, we identified all twin pairs for whom at least 1 twin later on was recorded under a stroke diagnosis in the death register. The first stroke death in a twin pair was defined as the index stroke death. The live co-twins were followed from the date of the index stroke death and until they were registered as dead as a result of stroke, other death causes, emigration, or end of study period, whichever event came first. Data were available through 1993, a total of up to 51 years of follow-up.
The National Discharge Registry was established in 1977 and comprises information on all discharges from Danish hospitals, including day of admission and discharge, hospital code, department code, and up to 20 discharge diagnoses. Data from both the discharge register and the death register were available for the period of 1977 through 1993. Among 11 598 same-sex twin pairs with known zygosity and with both twins alive on January 1, 1977, we identified all twin pairs for whom at least 1 twin later on was registered as dead as a result of stroke or having a discharge diagnosis of intracerebral hemorrhage, cerebral infarction, or unspecified stroke according to ICD-8 and ICD-10 described above. The first stroke admission or stroke death in a twin pair was defined as the index stroke event. The co-twins were followed from the date of the index stroke event and until they were admitted with a stroke diagnosis, registered dead as a result of stroke, or censored because of death resulting from other causes, emigration, or end of study period, whichever event came first.
We studied the validity of the stroke diagnoses from the National Discharge Registry in all twins discharged with a stroke diagnosis from hospitals in Funen County (465 000 inhabitants; 9% of the Danish population) within the period of 1977 through 1998. Data from discharge records were abstracted in a structured form and evaluated by a neurologist (S.B.). Stroke was defined according to the World Health Organization criteria.17 A review diagnosis of intracerebral hemorrhage was considered if the clinical signs were compatible with stroke and an intracerebral hemorrhage was found at the neuroradiological examination, surgery, or autopsy. Cases with clinical signs of stroke were classified as cerebral infarction if an intracerebral hemorrhage was excluded by means of auxiliary examinations. A total of 333 stroke discharge diagnoses were identified. Discharge abstracts from 1 medical department could not be retrieved, leaving a total of 288 stroke hospitalizations in 198 twins for the validation study. The positive predictive value of a register diagnosis (the review diagnosis being the gold standard) of stroke was 85% (95% CI, 79 to 90%). For a register diagnosis of intracerebral hemorrhage, the positive predictive value was only moderate (71%; 95% CI, 54 to 85), and the predictive value for cerebral infarction was low (58%; 95% CI, 46 to 70). These results reflected primarily the limited availability of neuroimaging in the study period; thus, a reliable stroke subtype classification was not possible.
We calculated the time from the index stroke death in a twin pair until stroke death or censoring in the co-twin. The risk of stroke death in monozygotic and dizygotic co-twins was compared by the log-rank test. A Cox proportional-hazards model was used to estimate the crude and age- and sex-adjusted relative risk and 95% CI of stroke death in monozygotic versus dizygotic co-twins. The assumptions of proportional hazards were tested by means of log-log plots.
For the period of 1977 through 1993, we calculated the time from the index stroke hospitalization or stroke death in a twin pair until stroke admission, stroke death, or censoring in the co-twin. The risk of stroke hospitalization or stroke death in monozygotic and dizygotic twins was compared by the log-rank test. Age- and sex-adjusted relative risk of stroke hospitalization or stroke death was estimated by means of a Cox proportional-hazards model.
To study the effect of calendar period and truncation, we entered index stroke death or index stroke admission as belonging to the first or second half of the observation periods, respectively, in the models.
Twin similarity for stroke was also assessed by calculation of pairwise and probandwise concordance rates for monozygotic and dizygotic twin pairs. The probandwise concordance rate is the proportion of affected co-twins among all individuals independently ascertained. This proportion equals the probability that the co-twin to a proband is affected, and it can be compared across studies even when the studies are characterized by different ascertainment probabilities.18 The concordance between monozygotic and dizygotic twin pairs was compared by χ2 test using the pairwise concordance rates, ie, the proportion of concordant pairs among all pairs affected.
Furthermore, calculation of concordance rates and hazard ratios for stroke death and stroke death or hospitalization in monozygotic and dizygotic twins was stratified by sex and age (<65 years, 65 to 74 years, and ≥75 years). The correlations for stroke were estimated in accordance with the multifactorial threshold model and expressed as tetrachoric correlations. Finally, we applied structural equation modeling to estimate the heritability of the liability to stroke.19,20⇓
We identified 990 same-sex twin pairs with both twins alive on January 1, 1943, and for whom at least 1 twin was recorded under a stroke diagnosis in the death register within the study period. In total, 559 of the co-twins were alive on the day of the index stroke death (Table 1). Median age at the beginning of follow-up was 72.6 years (range, 28.3 to 96.2 years). There was no significant difference between monozygotic and dizygotic co-twins concerning sex or age distribution. In all, 69 co-twins were registered as dead as a result of stroke within the study period. The median age of death of these co-twins was 81.9 years (range, 57.8 to 96.8 years). The crude risk of stroke death was 2.3 times higher in monozygotic compared with dizygotic co-twins (log-rank test, P<0.001). After adjustment for sex and age, monozygotic co-twins still had a double risk of stroke death compared with dizygotic co-twins (Table 2). Male sex and increasing age were also associated with increased risk of stroke death. Entering the calendar period of the index stroke death in the model had no effect on these results.
A total of 869 same-sex twin pairs were identified in which both twins were alive on January 1, 1977, and of which at least 1 twin later on was recorded under a stroke diagnosis in the discharge or death register. We excluded 186 co-twins who had died or emigrated before the index stroke event, leaving a cohort of 683 co-twins for further follow-up (Table 1). Median age at the beginning of follow-up was 69.4 years (range, 28.5 to 92.1 years). There were no significant sex or age distribution differences between monozygotic and dizygotic co-twins in this cohort. Within the study period, 72 co-twins were registered as either hospitalized or dead as a result of stroke. The median age of the co-twins on admission or stroke death was 78.2 years (range, 52.9 to 96.8 years). The crude risk of stroke admission or stroke death was 1.7 times higher in monozygotic compared with dizygotic co-twins (log-rank test, P=0.02). Monozygotic co-twins had a 1.5-times-higher risk of stroke hospitalization compared with dizygotic co-twins after adjustment for sex and age (Table 2). The difference was not statistically significant (P=0.11). Adjusting for calendar period of the index stroke admission had no effect on this result.
Among the 990 twin pairs with at least 1 twin registered with a stroke diagnosis in the death registry, stroke death was recorded in 5.0% of monozygotic twins and in 4.4% of dizygotic twins. The probandwise concordance rates were 0.18 for monozygotic pairs and 0.10 for dizygotic pairs (Table 3). The pairwise concordance rates (0.10 and 0.05 for monozygotic and dizygotic pairs, respectively) were significantly higher in monozygotic pairs compared with dizygotic pairs (χ2 test, P=0.006).
Probandwise and pairwise concordance rates were also calculated for all 869 twin pairs with at least 1 twin either hospitalized or dead as a result of stroke (Table 3). There was a trend toward a higher pairwise concordance rate in monozygotic compared with dizygotic pairs, and the difference was borderline statistically significant (χ2 test, P=0.06). Stroke hospitalization or stroke death was recorded in 4.6% of monozygotic and in 3.8% of dizygotic twins. These results did not materially change after stratification by sex or age group (results not shown).
The structural equation modeling to estimate heritability revealed that a model including additive genetic effects and shared and nonshared environmental effects provided the best fit to the data, both for stroke death alone and for stroke hospitalization and stroke death combined. The heritability estimate was 0.32 (95% CI, 0.04 to 0.47) for stroke death. Shared environmental effects contributed with 0.06 (95% CI, 0.00 to 0.27) and nonshared environmental effects contributed with 0.62 (95% CI, 0.53 to 0.73) of the liability to stroke death. For stroke hospitalization or death, the heritability estimate was 0.17 (95% CI, 0.00 to 0.44), whereas shared and nonshared environmental effects contributed with 0.24 (95% CI, 0.04 to 0.42) and 0.59 (95% CI, 0.49 to 0.69), respectively.
In the present study, we identified all twin pairs with at least 1 twin registered dead as a result of stroke and found that the risk of stroke death for monozygotic co-twins was increased 2-fold compared with dizygotic co-twins. Similarly, in twin pairs with at least 1 twin hospitalized or dead as a result of stroke, monozygotic co-twins had a 1.5-times-higher risk of stroke admission or stroke death compared with dizygotic co-twins. Heritability estimates revealed a moderate genetic influence on the liability to stroke death and stroke death and hospitalization combined. The major advantages of the present study were the application of a nationwide, population-based twin register and a long-term, complete follow-up of the co-twins. Discharge diagnoses and data on causes of death came from registries and thus were free of recall bias.
Previously, only 2 twin studies on stroke have been reported, and the results were inconsistent. A study of stroke death based on death certificates was carried out in a cohort of 10 900 Swedish twin pairs.7 The study showed no difference between concordance rates for stroke death in monozygotic compared with dizygotic twin pairs. However, the study comprised only 19 pairs concordant for stroke, and the follow-up period was only 12 years. Brass et al21 conducted a questionnaire study of 2722 twin pairs and found a 4-fold increase in the relative risk of stroke in monozygotic compared with dizygotic twin pairs. The probandwise concordance rates were 0.18 for monozygotic twins and 0.04 for dizygotic twins, which were comparable to the observed rates in our study. However, the twin cohort reported by Brass et al21 consisted of male army veterans only; stroke was self-reported and therefore was restricted to surviving stroke cases; and the study comprised only 8 pairs concordant for stroke. Heritability estimates of the liability to stroke were not reported in these twin studies.
The importance of a family history of stroke has been assessed in several studies. In a cohort of 2317 individuals from the Framingham Offspring Study, Kiely et al8 traced 34 cases of stroke or transient ischemic attack. Individuals with a parental history of stroke or transient ischemic attack were 1.5 times more likely to have stroke or transient ischemic attack than individuals without a positive parental history. Wannamethee et al9 traced 278 stroke cases in a cohort of 7735 middle-aged British men and found that the relative risk of stroke in men with a parental history of stroke death was 1.4 after adjustment for age and other stroke risk factors. In a cohort study of 14 371 middle-aged Finns, the relative risk of subarachnoid hemorrhage, intracerebral hemorrhage, and ischemic stroke associated with a parental history of stroke before 60 years of age was 1.9 in men and 1.8 in women.10 The relative risk of ischemic stroke separately was 1.5 in men and 1.8 in women. Liao et al11 studied the personal history of stroke in 3168 probands and 29 325 of their first-degree relatives. Probands with a parental history of stroke were 1.9 times more likely to have a stroke compared with probands with no family history of stroke. However, contrary to twin studies, it is not possible to decide from family studies if a familial aggregation of a disease is due to genetic factors, shared environment, or both.
Our study has several potential limitations. The primary limitation was the possibility of diagnostic misclassification, which is a major concern in many studies using register diagnoses. In the present study, we assessed the validity of a stroke diagnosis from the discharge register and found an acceptable positive predictive value of 85% with the review diagnosis from discharge records as gold standard. The validity of a stroke diagnosis from the death register has not been assessed. It is unlikely, however, that misclassification of stroke cases in the registries is dependent on zygosity, and misclassification will therefore contribute to a bias toward finding no difference between the 2 zygosity groups. Consequently, our estimates of the relative risk of stroke death and stroke admission in monozygotic compared with dizygotic co-twins are probably conservative.
Stroke is a heterogeneous disease comprising subarachnoid hemorrhage, intracerebral hemorrhage, and ischemic stroke. Each of these subgroups can be further divided according to etiological characteristics, and the genetic contribution to stroke risk may depend on stroke subtype.
Subarachnoid hemorrhage differs considerably from other stroke subtypes with regard to origin and was not included in the present study. The discharge register diagnoses of intracerebral hemorrhage and ischemic stroke had a low positive predictive value and were therefore inapplicable to further analysis. Consequently, we cannot determine from our data whether a modest genetic stroke risk is present for all types of stroke or a larger genetic component is confined to specific stroke subtypes, as has been indicated for survival after lobar hemorrhage.22 We speculate, however, on whether the higher relative risk for fatal stroke compared with fatal and nonfatal stroke could be due to a higher proportion of hemorrhagic strokes in the fatal stroke group. If this is the case, it might reflect a larger genetic component in the origin of intracerebral hemorrhage compared with ischemic stroke.
Finally, information on other risk factors for stroke was not available, and we cannot exclude the possibility of confounding as a result of an unequal distribution of other stroke risk factors between monozygotic and dizygotic co-twins. However, if cerebrovascular risk factors are generally more common in monozygotic than dizygotic twins, a higher overall mortality rate in monozygotic twins would be expected. Furthermore, we would expect stroke to be more common in monozygotic than in dizygotic twins. In a previous study, mortality from 6 to 90 years of age was found to be similar in monozygotic twins, dizygotic twins, and the general population, suggesting that diseases that are common causes of death are no more common in monozygotic than in dizygotic twins or singletons.6 Similarly, we found that the occurrence of stroke was of the same magnitude in monozygotic and dizygotic twins. Consequently, substantial differences in the distribution of cerebrovascular risk factors between monozygotic and dizygotic twins seem unlikely.
In conclusion, the results from this nationwide twin study with a long-term follow-up suggest that genetic factors increase the risk of stroke and that the size of this effect is moderate.
This study was supported by a National Institute of Aging research grant (NIA-PO1-AG08761) and the Danish National Research Foundation.
- Received August 24, 2001.
- Revision received November 21, 2001.
- Accepted November 22, 2001.
- ↵Vila N, Obach V, Revilla M, Oliva R, Chamorro A. α1-Antichymotrypsin gene polymorphism in patients with stroke. Stroke. 2000; 31: 2103–2105.
- ↵Reiner AP, Kumar PN, Schwartz SM, Longstreth WT, Pearce RM, Rosendaal FR, Psaty BM, Siscovick DS. Genetic variants of platelet glycoprotein receptors and risk of stroke in young women. Stroke. 2000; 31: 1628–1633.
- ↵Hiramoto M, Yoshida H, Imaizumi T, Yoshimizu N, Satoh K. A mutation in plasma platelet-activating factor acetylhydrolase (Val279→Phe) is a genetic risk factor for stroke. Stroke. 1997; 28: 2417–2420.
- ↵Elbaz A, Poirier O, Moulin T, Chedru F, Cambien F, Amarenco P. Association between the Glu298Asp polymorphism in the endothelial constitutive nitric oxide synthase gene and brain infarction. Stroke. 2000; 31: 1634–1639.
- ↵Sonoda A, Murata M, Ito D, Tanahashi N, Ohta A, Tada Y, Takeshita E, Yoshida T, Saito I, Yamamoto M, Ikeda Y, Fukuuchi Y, Watanabe K. Association between platelet glycoprotein Ιbα genotype and ischemic cerebrovascular disease. Stroke. 2000; 31: 493–497.
- ↵Christensen K, Vaupel JW, Holm NV, Yashin AI. Mortality among twins after age 6: fetal origins hypothesis versus twin method. BMJ. 1995; 310: 432–436.
- ↵Kiely DK, Wolf PA, Cupples A, Beiser AS, Myers RH. Familial aggregation of stroke: the Framingham study. Stroke. 1993; 24: 1366–1371.
- ↵Wannamethee SG, Shaper AG, Ebrahim S. History of parental death from stroke or heart trouble and the risk of stroke in middle-aged men. Stroke. 1996; 27: 1492–1498.
- ↵Jousilahti P, Rastenyte D, Tuomilehto J, Sarti C, Vartiainen E. Parental history of cardiovascular disease and risk of stroke. Stroke. 1997; 28: 1361–1366.
- ↵Liao D, Myers R, Hunt S, Shahar E, Paton C, Burke G, Province M, Heiss G. Familial history of stroke and stroke risk: the Family Heart Stroke. 1997; 28: 1908–1912.
- ↵Kubota M, Yamaura A, Ono J, Itani T, Tachi N, Ueda K, Nagata I, Sugimoto S. Is family history an independent risk factor for stroke? J Neurol Neurosurg Psychiatry. 1997; 62: 66–70.
- ↵Boysen G, Nyboe J, Appleyard M, Sørensen PS, Boas J, Somnier F, Jensen G, Schnohr P. Stroke incidence and risk factors for stroke in Copenhagen, Denmark. Stroke. 1988; 19: 1345–1353.
- ↵Graffagnino C, Gasecki AP, Doig GS, Hachinski VC. The importance of family history in cerebrovascular disease. Stroke. 1994; 25: 1599–1604.
- ↵Hauge M. The Danish Twin Register.In: Mednick SA, Baert AE, Bachmann BP, eds. Prospective Longitudinal Research. Oxford, UK: Oxford Medical; 1981: 217–221.
- ↵World Health Organization. Recommendations on stroke prevention, diagnosis, and therapy. Stroke. 1989; 20: 1407–1431.
- ↵McGue M. When assessing twin concordance, use the probandwise not the pairwise rate. Schizophr Bull. 1992; 18: 171–176.
- ↵Neale MC, Cardon LR. Methodology for Genetic Studies of Twins and Families. Dordrecht, Netherlands: Kluwer Academic Publishers; 1992.
- ↵Neale MC, Boker SM, Xie G, Maes HH. Mx: Statistical Modeling 5. Richmond, Va: University of Virginia, Department of Psychiatry; 1999.
- ↵Brass LM, Isaacsohn JL, Merikangas KR, Robinette CD. A study of twins and stroke. Stroke. 1992; 23: 221–223.