Abstract
The Mendelian randomization approach is an epidemiological study design incorporating genetic information into traditional epidemiological studies to infer causality of biomarkers, risk ...factors, or lifestyle factors on disease risk. Mendelian randomization studies often draw on novel information generated in genome-wide association studies on causal associations between genetic variants and a risk factor or lifestyle factor. Such information can then be used in a largely unconfounded study design free of reverse causation to understand if and how risk factors and lifestyle factors cause cardiovascular disease. If causation is demonstrated, an opportunity for prevention of disease is identified; importantly however, before prevention or treatment can be implemented, randomized intervention trials altering risk factor levels or improving deleterious lifestyle factors needs to document reductions in cardiovascular disease in a safe and side-effect sparse manner. Documentation of causality can also inform on potential drug targets, more likely to be successful than prior approaches often relying on animal or cell studies mainly. The present review summarizes the history and background of Mendelian randomization, the study design, assumptions for using the design, and the most common caveats, followed by a discussion on advantages and disadvantages of different types of Mendelian randomization studies using one or more samples and different levels of information on study participants. The review also provides an overview of results on many of the risk factors and lifestyle factors for cardiovascular disease examined to date using the Mendelian randomization study design.
Increased nonfasting remnant cholesterol, like increased LDL cholesterol, is causally associated with increased risk for ischemic heart disease (IHD). We tested the hypothesis that extreme ...concentrations of nonfasting remnant and LDL cholesterol are equal contributors to the risk of IHD, myocardial infarction (MI), and all-cause mortality.
We compared stepwise increasing concentrations of nonfasting remnant and LDL cholesterol for association with risk of IHD, MI, and all-cause mortality in approximately 90 000 individuals from the Danish general population. During up to 22 years of complete follow-up, 4435 participants developed IHD, 1722 developed MI, and 8121 died.
Compared with participants with nonfasting remnant cholesterol <0.5 mmol/L (19.3 mg/dL), hazard ratios for IHD ranged from 1.3 (95% CI 1.1-1.5) for remnant cholesterol of 0.5-0.99 mmol/L (19.3-38.2 mg/dL) to 2.4 (1.9-2.9) for remnant cholesterol of ≥1.5 mmol/L (58 mg/dL) (P for trend <0.001). Compared with participants with LDL cholesterol <3.0 mmol/L (115.8 mg/dL), hazard ratios for IHD ranged from 1.3 (1.1-1.5) for LDL cholesterol of 3-3.99 mmol/L (115.8-154 mg/dL) to 2.3 (1.9-2.8) for LDL cholesterol of ≥5 mmol/L (193 mg/dL) (P < 0.001). Corresponding hazard ratios for MI ranged from 1.8 (1.4-2.3) to 3.4 (2.5-4.8) for remnant cholesterol (P < 0.001), and from 1.7 (1.4-2.2) to 4.7 (3.5-6.3) for LDL cholesterol (P < 0.001). Nonfasting remnant cholesterol concentrations were associated stepwise with all-cause mortality ranging from hazard ratio 1.0 (0.9-1.1) to 1.6 (1.4-1.9) (P < 0.001), whereas LDL cholesterol concentrations were associated with decreased all-cause mortality risk in a U-shaped pattern, with hazard ratios from 0.8 (0.7-0.8) to 0.9 (0.8-1.0) (P = 0.002). After mutual adjustment, LDL cholesterol best predicted MI, and remnant cholesterol best predicted all-cause mortality.
Both lipoproteins were associated equally with risk of IHD and MI; however, only nonfasting remnant cholesterol concentrations were associated stepwise with increased all-cause mortality risk.
High Lp(a) (lipoproteina) cause cardiovascular disease (CVD) in a primary prevention setting; however, it is debated whether high Lp(a) lead to recurrent CVD events. We tested the latter hypothesis ...and estimated the Lp(a)-lowering needed for 5 years to reduce CVD events in a secondary prevention setting. Approach and Results: From the CGPS (Copenhagen General Population Study; 2003-2015) of 58 527 individuals with measurements of Lp(a), 2527 aged 20 to 79 with a history of CVD were studied. The primary end point was major adverse cardiovascular event (MACE). We also studied 1115 individuals with CVD from the CCHS (Copenhagen City Heart Study; 1991-1994) and the CIHDS (Copenhagen Ischemic Heart Disease Study; 1991-1993). During a median follow-up of 5 years (range, 0-13), 493 individuals (20%) experienced a MACE in the CGPS. MACE incidence rates per 1000 person-years were 29 (95% CI, 25-34) for individuals with Lp(a)<10 mg/dL, 35 (30-41) for 10 to 49 mg/dL, 42 (34-51) for 50 to 99 mg/dL, and 54 (42-70) for ≥100 mg/dL. Compared with individuals with Lp(a)<10 mg/dL (18 nmol/L), the multifactorially adjusted MACE incidence rate ratios were 1.28 (95% CI, 1.03-1.58) for 10 to 49 mg/dL (18-104 nmol/L), 1.44 (1.12-1.85) for 50 to 99 mg/dL (105-213 nmol/L), and 2.14 (1.57-2.92) for ≥100 mg/dL (214 nmol/L). Independent confirmation was obtained in individuals from the CCHS and CIHDS. To achieve 20% and 40% MACE risk reduction in secondary prevention, we estimated that plasma Lp(a) should be lowered by 50 mg/dL (95% CI, 27-138; 105 nmol/L 55-297) and 99 mg/dL (95% CI, 54-273; 212 nmol/L 114-592) for 5 years.
High concentrations of Lp(a) are associated with high risk of recurrent CVD in individuals from the general population. This study suggests that Lp(a)-lowering by 50 mg/dL (105 nmol/L) short-term (ie, 5 years) may reduce CVD by 20% in a secondary prevention setting.
Short telomeres in peripheral blood leukocytes are associated with older age and age-related diseases. We tested the hypotheses that short telomeres are associated with both increased cancer ...mortality and all-cause mortality.
Individuals (n = 64637) were recruited from 1991 onwards from two Danish prospective cohort studies: the Copenhagen City Heart Study and the Copenhagen General Population Study. All had telomere length measured by quantitative polymerase chain reaction and the genotypes rs1317082 (TERC), rs7726159 (TERT), and rs2487999 (OBFC1) determined. The sum of telomere-shortening alleles from these three genotypes was calculated. We conducted Cox regression analyses and instrumental variable analyses using the allele sum as an instrument. All statistical tests were two-sided.
Among 7607 individuals who died during follow-up (0-22 years, median = 7 years), 2420 had cancer and 2633 had cardiovascular disease as causes of death. Decreasing telomere length deciles were associated with increasing all-cause mortality (P(trend) = 2*10(-15)). The multivariable-adjusted hazard ratio of all-cause mortality was 1.40 (95% confidence interval CI = 1.25 to 1.57) for individuals in the shortest vs the longest decile. Results were similar for cancer mortality and cardiovascular mortality. Telomere length decreased 69 base pairs (95% CI = 61 to 76) per allele for the allele sum, and the per-allele hazard ratio for cancer mortality was 0.95 (95% CI = 0.91 to 0.99). Allele sum was not associated with cardiovascular, other, or all-cause mortality.
Short telomeres in peripheral blood leukocytes were associated with high mortality in association analyses. In contrast, genetically determined short telomeres were associated with low cancer mortality but not with all-cause mortality.
Exacerbations of respiratory symptoms in chronic obstructive pulmonary disease (COPD) have profound and long-lasting adverse effects on patients.
To test the hypothesis that elevated levels of ...inflammatory biomarkers in individuals with stable COPD are associated with an increased risk of having exacerbations.
Prospective cohort study examining 61,650 participants with spirometry measurements from the Copenhagen City Heart Study (2001-2003) and the Copenhagen General Population Study (2003-2008). Of these, 6574 had COPD, defined as a ratio between forced expiratory volume in 1 second (FEV1) and forced vital capacity below 0.7.
Baseline levels of C-reactive protein (CRP) and fibrinogen and leukocyte count were measured in participants at a time when they were not experiencing symptoms of exacerbations. Exacerbations were recorded and defined as short-course treatment with oral corticosteroids alone or in combination with an antibiotic or as a hospital admission due to COPD. Levels of CRP and fibrinogen and leukocyte count were defined as high or low according to cut points of 3 mg/L, 14 μmol/L, and 9 ×10(9)/L, respectively.
During follow-up, 3083 exacerbations were recorded (mean, 0.5/participant). In the first year of follow-up, multivariable-adjusted odds ratios for having frequent exacerbations were 1.2 (95% CI, 0.7-2.2; 17 events/1000 person-years) for individuals with 1 high biomarker, 1.7 (95% CI, 0.9-3.2; 32 events/1000 person-years) for individuals with 2 high biomarkers, and 3.7 (95% CI, 1.9-7.4; 81 events/1000 person-years) for individuals with 3 high biomarkers compared with individuals who had no elevated biomarkers (9 events/1000 person-years; trend: P = 2 × 10(-5)). Corresponding hazard ratios using maximum follow-up time were 1.4 (95% CI, 1.1-1.8), 1.6 (95% CI, 1.3-2.2), and 2.5 (95% CI, 1.8-3.4), respectively (trend: P = 1 × 10(-8)). The addition of inflammatory biomarkers to a basic model including age, sex, FEV1 percent predicted, smoking, use of any inhaled medication, body mass index, history of previous exacerbations, and time since most recent prior exacerbation improved the C statistics from 0.71 to 0.73 (comparison: P = 9 × 10(-5)). Relative risks were consistent in those with milder COPD, in those with no history of frequent exacerbations, and in the 2 studies separately. The highest 5-year absolute risks of having frequent exacerbations in those with 3 high biomarkers (vs no high biomarkers) were 62% (vs 24%) for those with Global Initiative for Chronic Obstructive Lung Disease (GOLD) grades C-D (n = 558), 98% (vs 64%) in those with a history of frequent exacerbations (n = 127), and 52% (vs 15%) for those with GOLD grades 3-4 (n = 465).
Simultaneously elevated levels of CRP and fibrinogen and leukocyte count in individuals with COPD were associated with increased risk of having exacerbations, even in those with milder COPD and in those without previous exacerbations. Further investigation is needed to determine the clinical value of these biomarkers for risk stratification.
Objectives The purpose of this study was to determine whether elevated lipoprotein(a) levels and corresponding LPA risk genotypes (rs10455872, rs3798220, kringle IV type 2 repeat polymorphism) ...prospectively associate with increased risk of aortic valve stenosis (AVS). Background The etiologic basis of AVS is unclear. Recent data implicate an LPA genetic variant (rs10455872), associated with Lp(a) levels, in calcific AVS. Methods We combined data from 2 prospective general population studies, the Copenhagen City Heart Study (1991 to 2011; n = 10,803) and the Copenhagen General Population Study (2003 to 2011; n = 66,877), following up 77,680 Danish participants for as long as 20 years, during which time 454 were diagnosed with AVS. We conducted observational and genetic instrumental variable analyses in a Mendelian randomization study design. Results Elevated Lp(a) levels were associated with multivariable adjusted hazard ratios for AVS of 1.2 (95% confidence interval CI: 0.8 to 1.7) for 22nd to 66th percentile levels (5 to 19 mg/dl), 1.6 (95% CI: 1.1 to 2.4) for 67th to 89th percentile levels (20 to 64 mg/dl), 2.0 (95% CI: 1.2 to 3.4) for 90th to 95th percentile levels (65 to 90 mg/dl), and 2.9 (95% CI: 1.8 to 4.9) for levels greater than 95th percentile (>90 mg/dl), versus levels less than the 22nd percentile (<5 mg/dl; trend, p < 0.001). Lp(a) levels were elevated among carriers of rs10455872 and rs3798220 minor alleles, and of low number of KIV-2 repeats (trend, all p < 0.001). Combining all genotypes, instrumental variable analysis yielded a genetic relative risk for AVS of 1.6 (95% CI: 1.2 to 2.1) for a 10-fold Lp(a) increase, comparable to the observational hazard ratio of 1.4 (95% CI: 1.2 to 1.7) for a 10-fold increase in Lp(a) plasma levels. Conclusions Elevated Lp(a) levels and corresponding genotypes were associated with increased risk of AVS in the general population, with levels >90 mg/dl predicting a threefold increased risk.
Lipid profiles are usually measured after fasting. We tested the hypotheses that these levels change only minimally in response to normal food intake and that nonfasting levels predict cardiovascular ...events.
We cross-sectionally studied 33 391 individuals 20 to 95 years of age from the Copenhagen General Population Study. We also studied 9319 individuals 20 to 93 years of age from the Copenhagen City Heart Study, 1166 of whom developed cardiovascular events during 14 years of follow-up. Compared with fasting levels, total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein (HDL) cholesterol, and albumin levels were reduced up to 3 to 5 hours after the last meal; triglycerides levels were increased up to 6 hours after the last meal; and non-HDL cholesterol level, apolipoprotein A1 level, apolipoprotein B level, ratio of total cholesterol to HDL cholesterol, and ratio of apolipoprotein B to apolipoprotein A1 did not change in response to normal food intake. The maximum changes after normal food and fluid intake from fasting levels were -0.2 mmol/L for total cholesterol, -0.2 mmol/L for low-density lipoprotein cholesterol, -0.1 mmol/L for HDL cholesterol, and 0.3 mmol/L for triglycerides. Highest versus lowest tertile of nonfasting total cholesterol, non-HDL cholesterol, low-density lipoprotein cholesterol, apolipoprotein B, triglycerides, ratio of total cholesterol to HDL cholesterol, and ratio of apolipoprotein B/apolipoprotein A1 and lowest versus highest tertile of nonfasting HDL cholesterol and apolipoprotein A1 predicted 1.7- to 2.4-fold increased risk of cardiovascular events.
Lipid profiles at most change minimally in response to normal food intake in individuals in the general population. Furthermore, nonfasting lipid profiles predicted increased risk of cardiovascular events.
Plasma apolipoprotein B (apoB) is a composite measure of all apoB-containing lipoproteins causing atherosclerotic cardiovascular disease; however, it is unclear which fraction of risk is explained by ...cholesterol and triglycerides, respectively, in very low-density lipoproteins (VLDLs).
The authors tested the hypothesis that VLDL cholesterol and triglycerides each explain part of the myocardial infarction risk from apoB-containing lipoproteins.
Nested within 109,751 individuals from the Copenhagen General Population Study, the authors examined 25,480 subjects free of lipid-lowering therapy and myocardial infarction at study entry. All had measurements of plasma apoB (quantitating number of apoB-containing lipoproteins) and cholesterol and triglyceride content of VLDL, intermediate-density lipoproteins (IDLs), and low-density lipoproteins (LDLs).
During a median 11 years of follow-up, 1,816 were diagnosed with myocardial infarction. Per 1-mmol/l higher levels, multivariable-adjusted hazard ratios for myocardial infarction were 2.07 (95% confidence interval CI: 1.81 to 2.36) for VLDL cholesterol, 1.19 (95% CI: 1.14 to 1.25) for VLDL triglycerides, 5.38 (95% CI: 3.73 to 7.75) for IDL cholesterol, and 1.86 (95% CI: 1.62 to 2.14) for LDL cholesterol. Per 1-g/l higher plasma apoB, the corresponding value was 2.21 (95% CI: 1.90 to 2.58). In a step-up Cox regression, risk factors for myocardial infarction entered by importance as VLDL cholesterol, systolic blood pressure, smoking, and IDL + LDL cholesterol, whereas VLDL triglycerides did not enter the model. VLDL cholesterol explained 50% and IDL + LDL cholesterol 29% of the risk of myocardial infarction from apoB-containing lipoproteins, whereas VLDL triglycerides did not explain risk.
VLDL cholesterol explained one-half of the myocardial infarction risk from elevated apoB-containing lipoproteins, whereas VLDL triglycerides did not explain risk.
Display omitted
Abstract
Aims
In observational studies, non-alcoholic fatty liver disease (NAFLD) is associated with high risk of ischaemic heart disease (IHD). We tested the hypothesis that a high liver fat content ...or a diagnosis of NAFLD is a causal risk factor for IHD.
Methods and results
In a cohort study of the Danish general population (n = 94 708/IHD = 10 897), we first tested whether a high liver fat content or a diagnosis of NAFLD was associated observationally with IHD. Subsequently, using Mendelian randomization, we tested whether a genetic variant in the gene encoding the protein patatin-like phospholipase domain containing 3 protein (PNPLA3), I148M (rs738409), a strong and specific cause of high liver fat content and NAFLD, was causally associated with the risk of IHD. We found that the risk of IHD increased stepwise with increasing liver fat content (in quartiles) up to an odds ratio (OR) of 2.41 (1.28–4.51)(P-trend = 0.004). The corresponding OR for IHD in individuals with vs. without NAFLD was 1.65 (1.34–2.04)(P = 3×10−6). PNPLA3 I148M was associated with a stepwise increase in liver fat content of up to 28% in MM vs. II-homozygotes (P-trend = 0.0001) and with ORs of 2.03 (1.52–2.70) for NAFLD (P = 3×10−7), 3.28 (2.37–4.54) for cirrhosis (P = 4×10−12), and 0.95 (0.86–1.04) for IHD (P = 0.46). In agreement, in meta-analysis (N = 279 013/IHD = 71 698), the OR for IHD was 0.98 (0.96–1.00) per M-allele vs. I-allele. The OR for IHD per M-allele higher genetically determined liver fat content was 0.98 (0.94–1.03) vs. an observational estimate of 1.05 (1.02–1.09)(P for comparison = 0.02).
Conclusion
Despite confirming the known observational association of liver fat content and NAFLD with IHD, lifelong, genetically high liver fat content was not causally associated with risk of IHD. These results suggest that the observational association is due to confounding or reverse causation.
Abstract
Context:
Recently, data on 2,000,000 people established that low body mass index (BMI) is associated with increased risk of dementia. Whether this observational association reflects a causal ...effect remains to be clarified.
Objective:
We tested the hypothesis that there is a causal association between low BMI and high risk of Alzheimer’s disease.
Design, Setting, and Participants:
Using a Mendelian randomization approach, we studied 95,578 individuals from the Copenhagen General Population Study (CGPS) with up to 36 years of follow-up and consortia data on 303,958 individuals from the Genetic Investigation of Anthropometric Traits (GIANT) and the International Genomics of Alzheimer's Project (IGAP).
Main Outcome Measure:
Risk of Alzheimer’s disease.
Results:
The causal odds ratio for a 1-kg/m2 genetically determined lower BMI was 0.98 95% confidence interval (CI), 0.77 to 1.23 for a weighted allele score in the CGPS. Using 32 BMI-decreasing variants from GIANT and IGAP the causal odds ratio for Alzheimer’s disease for a 1-standard deviation (SD) lower genetically determined BMI was 1.02 (95% CI, 0.86 to 1.22). Corresponding observational hazard ratios from the CGPS were 1.07 (95% CI, 1.05 to 1.09) and 1.32 (95% CI, 1.20 to 1.46) for a 1-kg/m2 and a 1-SD lower BMI, respectively.
Conclusions:
Genetic and hence lifelong low BMI is not associated with increased risk of Alzheimer’s disease in the general population. These data suggest that low BMI is not a causal risk factor for Alzheimer’s disease and that the corresponding observational association likely is explained by reverse causation or confounding.
In this Mendelian randomization study of 399,536 individuals, genetic and hence lifelong lower BMI was not associated with higher risk of Alzheimer’s disease, in contrast to observational associations.