Family history, comorbidity and risk of thoracic aortic disease: a population-based case-control study
- 1Department of Molecular Medicine and Surgery, Cardiovascular Surgery Unit, Karolinska Institutet, Stockholm, Sweden
- 2Department of Medicine, Clinical Epidemiology Unit, Karolinska Institutet, Stockholm, Sweden
- 3Department of Surgical Sciences, Cardiovascular Surgery Unit, Uppsala University, Uppsala, Sweden
- Correspondence to Christian Olsson, Department of Molecular Medicine and Surgery, Cardiovascular Surgery Unit, Karolinska Universitetssjukhuset, Thoraxkliniken, Stockholm SE17176, Sweden;
- Received 16 January 2013
- Accepted 4 April 2013
- Published Online First 27 April 2013
Objective To examine the risk of thoracic aortic disease (TAD) when one or more first-degree relatives are affected, and to relate the risk of family history to the risk of other cardiopulmonary comorbidity.
Design Population-based, matched, case-control study.
Setting Registry-based investigation.
Patients All cases, nationwide, of TAD diagnosed 2001–2005 in individuals born 1932 or later (n=2436) were identified, and a random control-group (n=12 152) matched for age, sex and geography was generated. First-degree relatives were identified in the Multigeneration Registry. Family history of TAD was assessed by cross-linking nationwide health registries.
Results Family history was present in 108 cases (4.4%), compared with 93 (0.77%) controls (p<0.0001). The risk of TAD increased with number of affected relatives: OR 5.8 (95% CI 4.3 to 7.7) vs OR 20 (2.2 to 179) with one versus two or more affected relatives. The relative risk of TAD was highest in the youngest (≤49 years) age group and slightly more pronounced in women than in men (OR 7.2 (4.2 to 12) vs OR 5.5 (3.9 to 7.7)). Among cardiopulmonary comorbidities, heart failure conferred the highest relative risk, OR 6.3 (4.1 to 9.8).
Conclusions Family history confers a significantly increased (sixfold to 20-fold) relative risk of TAD. The effect is more pronounced in women and in younger subjects, and is not conveyed by cardiopulmonary comorbidity. Knowledge of family history is important to counselling, treatment indications, surveillance and screening protocols.
Familial aggregation of thoracic aortic diseases (TAD)—predominantly aneurysms and dissections—has been established in case series and cohorts derived from surgically treated patients.1–4 In such studies, family history is demonstrable in 14–20% of cases, promoting its role as an important risk factor for TAD. Family history of TAD has been associated with diagnosis at younger age and faster aortic growth, increasing sudden rupture risk and introducing uncertainty of adequate surgical timing.2–4 Family history has also been implied in late treatment failure (pseudoaneurysm or adjacent aneurysm formation requiring reintervention) in abdominal aortic aneurysm (AAA), suggesting progressive development of aortic disease and, eventually, worse prognosis.5 Knowledge on the magnitude of risk conveyed by family history, and its relation to other potential risk factors, is relevant to patient and family counselling, but also of importance in the planning of surveillance and definition of treatment indications. Acknowledging an increased burden of TAD, consideration must also be given to developing screening programmes, could adequate high-risk subgroups be identified.6 ,7 To accurately describe the impact of family history on the risk of TAD, population-based epidemiological data are needed. Based on Swedish nationwide registries of first-degree relatives, in-patient care and cause of death, a matched case-control study was designed to investigate the risk of developing thoracic aortic aneurysm or dissection associated with family history and adjusted for common cardiopulmonary risk factors.
Swedish citizens are issued a 12-digit unique personal identification number at birth, and so are permanent residents, allowing their identification in registries. The National Hospital Discharge Registry (HDR) and the Cause of Death Registry (CODR), maintained by The National Board of Health and Welfare, were used to identify all cases of thoracic aortic aneurysm (including thoracoabdominal, but excluding abdominal aneurysms) or dissection based on the relevant diagnostic codes in the Swedish adaptations of the 9–10th revisions of The International Classification of Diseases (ICD-9, ICD-10), table 1. Aneurysms of non-specified locations were not included. When adequately coded (table 1), traumatic aortic injuries, mycotic aneurysms and inflammatory aortic disease (ie, TAD without family history implications) were not included. Controls were randomly selected from the Total Population Registry (Statistics Sweden). The family history of the index cases and controls was obtained by identifying all first-degree relatives in the Multigeneration Registry (Statistics Sweden)—a registry including first-degree relatives (parents, children, siblings and half-siblings) of subjects born 1932 or later—and identifying diagnoses of thoracic or AAA in these subjects, if any. Besides age and sex, atherosclerosis, hypertension and smoking are implied risk factors for TAD.8 Their influence on the risk of TAD was assessed in terms of occurrence of correlated clinical conditions identifiable in discharge records; ischaemic heart disease, congestive heart failure, hypertension and chronic obstructive pulmonary disease. To minimise the chance of simple coexistence of conditions, the putative risk factors had to present prior (up to 5 years) to the initial TAD diagnosis. The study was approved by the regional research ethics committee, waiving individual informed consent.
Index cases (n=2436) were individuals with TAD according to ICD-10 and defined as a first occurrence in the HDR or as cause of death (direct or contributing or underlying) in the CODR during the period 2000–2005. To each index case, five subjects, alive and without known TAD at the index date of the corresponding case, were randomly selected. Controls (n=12 152) were matched to have the same sex and age as the index case, and to being residents of the same geographic area, forming a total study population of 14 588.
The matched case-control model was analysed with conditional logistic regression analyses. Results are presented as ORs with Wald 95% CIs. The risk of TAD was assessed in relation to family history of TAD categorised according to the number of affected first-degree relatives, family history of AAA (yes/no), and presence of the studied comorbid conditions (yes/no for each condition, as outlined above). Deviation from multiplicative effects of family history of TAD and other risk factors was assessed through introducing appropriate interaction terms into the model. Heterogeneity of the risk related to family history of TAD was assessed by Wald's testing. In assessing interactions, family history of TAD and comorbid conditions, respectively, were dichotomised. To display effects of combinations of different risk factors, compound variables were created from the dichotomised risk factor variables. The prevalence of risk factors among cases and controls was compared using χ2 test, and p values<0.05 were considered statistically significant. Statistical analyses were performed using SAS V.9.2 (SAS Institute, Cary, North Carolina, USA).
Of 2436 identified cases with TAD, 526 (21.6%) were identified postmortem in the CODR. A diagnosis of thoracic aortic aneurysm was established in 1204 cases (49.4%), thoracoabdominal aneurysm in 129 (5.3%), and aortic dissection in 1103 (45.3%). The male:female ratio was 2.6:1 (table 2). A TAD diagnosis was most common in the 60–69 years age group. In all, 108 cases (4.4%) had a positive family history as compared with 93 (0.77%) of controls, p<0.0001. Comorbidity, predominantly hypertension, was more common among cases, 542 (23%) versus 944 (7.8%), p<0.0001. Marfan syndrome was confirmed in 28 (1.1%) cases versus none in the control group, p<0.0001.
The risk of TAD with positive family history varied markedly with age: OR 34 (95% CI 12 to 98) in individuals up to 49 years; OR 4.8 (3.3 to 7.0) in 50–64-year-olds and OR 4.1 (2.4 to 7.0) in individuals 65 years or older. Family history of TAD increased the relative risk (OR 5.9) and further (OR 20) with more than one affected first-degree relative (table 3). The increase in relative risk was lower in males (OR 5.5 (3.9 to 7.7)) than in females (OR 7.2 (4.2 to 12)). A family history of AAA was associated with a slightly increased (OR 1.8) risk of TAD. The risk of TAD was modulated by adjusting for comorbidity; the strongest association of TAD and individual comorbid conditions was found for heart failure, table 3. When adjusting for all variables in the final statistical model, the effects were essentially unaltered (table 3 Model 2). Testing for interactions between family history of TAD and age, sex, AAA and comorbidity, only age was statistically significant (p=0.002).
The effects of family history of TAD, family history of AAA, and comorbidity were demonstrably close to multiplicative, also displaying the risks of combinations of risk factors (table 4). Too few subjects had a combination of all three risk factors to allow analysis. The combination of positive family history of TAD and any comorbidity entailed the highest relative risk, OR 15.
This is the first report on the relative risk of family history in TAD based on a large, unselected, nationwide, population-based sample forming a matched case-control study. Having a first-degree relative affected by TAD is associated with a sixfold increased risk of TAD; with two affected relatives, the risk is increased 20-fold. The impact of family history is highest in the youngest (≤49 years) age group, and slightly more pronounced in women. Family history of TAD displays a stronger association with TAD than several other variables, including family history of AAA and cardiovascular and pulmonary conditions.
Biddinger et al1 were among the first to describe the familial aggregation of TAD. Comparing the prevalence of TAD in 843 first-degree relatives of 158 patients referred for surgery with 547 controls, they found TAD in 17 of the proband relatives versus two in controls (p=0.0096), and calculated a relative risk of TAD of 1.8 for fathers and sisters and 10.9 for brothers. They also noted that, although thought to have different aetiologies, analysing aneurysms and dissections separately did not make a statistically significant contribution toward predicting first-degree relative aortic disease, supporting the use of TAD as an appropriate entity in epidemiological studies. In a series of important contributions, the Yale group reported that 19.3–21.5% of patients referred for surgery had a positive family history, and that those patients were younger (57 vs 64 years) and exhibited increased aortic growth rates (0.22 vs 0.03 cm/year) compared with sporadic cases.2–3 Though comprehensive, these studies were limited in size and again from surgical patient cohorts, introducing selection bias; individuals scheduled for surgery (or operated acutely) would represent a severely affected minority of individuals harbouring, or prone to develop, TAD as well as ignoring cases diagnosed posthumously.
In Sweden, combining unique personal identification numbers, validated nationwide registries with full coverage, and high autopsy rates, conditions are well suited for epidemiological studies, generating results generalisable at least to similarly homogeneous Caucasian populations. Two studies using similar methods have been undertaken previously. Hemminki et al8 in a study utilising the same registries for an earlier study era (1987–2001), but restricted to 71 identified sib-pairs, reported an overall standardised incidence ratio of 8.71 for individuals with an affected sibling. They, however, included all types of aneurysms in their study, noting that, since AAA constituted almost 40% and aneurysms with unspecified locations 6.7%, TAD constituted approximately 50% of the study population. Furthermore, only subjects up to 69 years of age were included. Recently, Larsson and coworkers presented epidemiological data for family history impact in AAA disease.9 They identified 3183 cases with AAA and matched them to 15 943 randomly selected controls and found an overall substantially lower relative risk of AAA of 1.9 with a positive family history, increasing to 4.7 with two affected relatives and to 10.0 with three affected relatives. Exercising appropriate caution with direct comparisons, since the populations and methods were highly similar, it seems reasonable to suggest that family history is relatively more important in TAD than in AAA.
Clearly, having a first-degree relative with TAD is associated with an increased risk of harbouring or developing TAD. The magnitude of relative risk increase is probably higher than previously reported, herein found to be 6–20 depending on the number of affected relatives. This may be an underestimation because of the inclusion of subjects up to a maximum of 73 years of age, based on the ramifications of the Multigeneration Registry. It must also be pointed out that the increase in absolute risk of TAD is limited; only 4.4% of cases (n=108) had a positive family history of TAD (comparing with a previously reported 3.1–3.2%).1 ,3 Marfan syndrome was not separately analysed due to too few identified cases, which at least, in part, must be caused by erroneous omission of the applicable ICD code (table 1) in these subjects. Other misclassifications and erroneous diagnoses will appear in large registries, but can be assumed to be evenly distributed, and not by themselves introducing bias or affecting the magnitude of relative effects. The constituting conditions of TAD (aneurysms, dissections and other subtypes) have not been analysed separately, and the impact of family history may vary between these conditions, as suggested in genetic studies.10 In absolute terms, cardiopulmonary comorbid conditions remain important as risk factors, prevalent in 25% in the present study albeit not conveying a risk increase of the same magnitude as family history. The effects of comorbidity were not altered by adjusting for family history, implying that the effect of family history is not mediated by inheriting the comorbid condition(s) such as hypertension. Heart failure was the comorbidity with the strongest statistical association to TAD. Heart failure may represent symptomatic left ventricular hypertrophy, or even end-stage hypertensive disease, acting as a marker of severe and/or untreated hypertension. Alternatively, it may be associated with aortic regurgitation due to aortic root dilation, possibly entailing left ventricular dilation and failure.
TAD fits appropriateness criteria for screening, being a lethal disease with a substantial asymptomatic preclinical phase benefiting from early, elective treatment and, arguably, with sufficient prevalence in selected groups, as demonstrated in the case of positive family history. As previously reported, as many as 22% of TAD cases are diagnosed postmortem, suggesting that overall mortality rate could be improved by earlier detection and treatment.6 Our findings lend support to the recommendation in the recently published Guidelines on TAD of screening of asymptomatic first-degree relatives every second year.11 It remains to be proved that subjects with familial TAD benefit from earlier surgical or endovascular intervention in terms of lower incidence of acute dissection and rupture, and reduced aortic-related mortality.
Apart from Marfan syndrome, other uncommon genetic conditions (Ehlers–Danlos type IV, Loeys–Dietz) with high prevalence of TAD lack specific ICD codes and do not lend themselves to registry-based study. Individuals with such conditions, not representative of cases with family history in general, may appear among cases. The limited number of cases with several affected first-degree relatives does not suggest an undue overrepresentation. The hospital discharge and CODRs employed in the study approach complete nationwide coverage but are primarily administrative databases. As such, they include all relevant diagnoses regardless of symptoms or interventions. Their data reliability and validity have been demonstrated repeatedly.12 ,13 Autopsy rates have declined in Sweden and demonstrate regional and local variability.14 At approximately 30%, they still remain substantially higher than a quoted <5% in North America.15 Meanwhile, diagnostic chest CT is used increasingly, in emergency departments as well as portmortem, gradually supplanting the need to verify presence of TAD by autopsy.16
The relative risk of TAD increases approximately sixfold to 20-fold with family history. The increase is more pronounced with more than one affected first-degree relative, in younger individuals, and in women. In comparison with AAA, family history may convey a larger relative risk in TAD. Cardiovascular and pulmonary conditions are commonly associated with TAD, and add further to but do not themselves convey the increased risk. Family history of TAD is important in counselling and medical decision making including screening, surveillance and intervention.
Contributors CO: Study design and conception, manuscript drafting and writing. FG: Study design and conception, data acquisition and analysis, manuscript review and approval. ES: Study design and conception, manuscript review and approval. Overall responsibility.
Funding The study was supported by The Faculty of Medicine and Pharmacy, Uppsala University, Uppsala, Sweden.
Competing interests None.
Ethics approval The regional research ethics committee, Uppsala, Sweden.
Provenance and peer review Not commissioned; externally peer reviewed.