Prioritising genetic findings for drug target identification and validation

The decreasing costs of high-throughput genetic sequencing and increasing abundance of sequenced genome data have paved the way for the use of genetic data in identifying and validating potential drug targets. However, the number of identified potential drug targets is often prohibitively large to experimentally evaluate in wet lab experiments, highlighting the need for systematic approaches for target prioritisation. In this review, we discuss principles of genetically guided drug development, specifically addressing loss-of-function analysis, colocalization and Mendelian randomisation (MR), and the contexts in which each may be most suitable. We subsequently present a range of biomedical resources which can be used to annotate and prioritise disease-associated proteins identified by these studies including 1) ontologies to map genes, proteins, and disease, 2) resources for determining the druggability of a potential target, 3) tissue and cell expression of the gene encoding the potential target, and 4) key biological pathways involving the potential target. We illustrate these concepts through a worked example, identifying a prioritised set of plasma proteins associated with non-alcoholic fatty liver disease (NAFLD). We identified five proteins with strong genetic support for involvement with NAFLD: CYB5A, NT5C, NCAN, TGFBI and DAPK2. All of the identified proteins were expressed in both liver and adipose tissues, with TGFBI and DAPK2 being potentially druggable. In conclusion, the current review provides an overview of genetic evidence for drug target identification, and how biomedical databases can be used to provide actionable prioritisation, fully informing downstream experimental validation.

Penetrance and Prognosis of MYH7 Variant-Associated Cardiomyopathies: Results From a Dutch Multicenter Cohort Study

BACKGROUND

MYH7 variants cause hypertrophic cardiomyopathy (HCM), noncompaction cardiomyopathy (NCCM), and dilated cardiomyopathy (DCM). Screening of relatives of patients with genetic cardiomyopathy is recommended from 10 to 12 years of age onward, irrespective of the affected gene.

OBJECTIVES

This study sought to study the penetrance and prognosis of MYH7 variant-associated cardiomyopathies.

METHODS

In this multicenter cohort study, penetrance and major cardiomyopathy-related events (MCEs) were assessed in carriers of (likely) pathogenic MYH7 variants by using Kaplan-Meier curves and log-rank tests. Prognostic factors were evaluated using Cox regression with time-dependent coefficients.

RESULTS

In total, 581 subjects (30.1% index patients, 48.4% male, median age 37.0 years [IQR: 19.5-50.2 years]) were included. HCM was diagnosed in 226 subjects, NCCM in 70, and DCM in 55. Early penetrance and MCEs (age <12 years) were common among NCCM-associated variant carriers (21.2% and 12.0%, respectively) and DCM-associated variant carriers (15.3% and 10.0%, respectively), compared with HCM-associated variant carriers (2.9% and 2.1%, respectively). Penetrance was significantly increased in carriers of converter region variants (adjusted HR: 1.87; 95% CI: 1.15-3.04; P = 0.012) and at age ≤1 year in NCCM-associated or DCM-associated variant carriers (adjusted HR: 21.17; 95% CI: 4.81-93.20; P < 0.001) and subjects with a family history of early MCEs (adjusted HR: 2.45; 95% CI: 1.09-5.50; P = 0.030). The risk of MCE was increased in subjects with a family history of early MCEs (adjusted HR: 1.82; 95% CI: 1.15-2.87; P = 0.010) and at age ≤5 years in NCCM-associated or DCM-associated variant carriers (adjusted HR: 38.82; 95% CI: 5.16-291.88; P < 0.001).

CONCLUSIONS

MYH7 variants can cause cardiomyopathies and MCEs at a young age. Screening at younger ages may be warranted, particularly in carriers of NCCM- or DCM-associated variants and/or with a family history of MCEs at <12 years.

Circulating Acylcarnitines Associated with Hypertrophic Cardiomyopathy Severity: an Exploratory Cross-Sectional Study in MYBPC3 Founder Variant Carriers

Hypertrophic cardiomyopathy (HCM) is a relatively common genetic heart disease characterised by myocardial hypertrophy. HCM can cause outflow tract obstruction, sudden cardiac death and heart failure, but severity is highly variable. In this exploratory cross-sectional study, circulating acylcarnitines were assessed as potential biomarkers in 124 MYBPC3 founder variant carriers (59 with severe HCM, 26 with mild HCM and 39 phenotype-negative [G + P-]). Elastic net logistic regression identified eight acylcarnitines associated with HCM severity. C3, C4, C6-DC, C8:1, C16, C18 and C18:2 were significantly increased in severe HCM compared to G + P-, and C3, C6-DC, C8:1 and C18 in mild HCM compared to G + P-. In multivariable linear regression, C6-DC and C8:1 correlated to log-transformed maximum wall thickness (coefficient 5.01, p = 0.005 and coefficient 0.803, p = 0.007, respectively), and C6-DC to log-transformed ejection fraction (coefficient -2.50, p = 0.004). Acylcarnitines seem promising biomarkers for HCM severity, however prospective studies are required to determine their prognostic value.

Identifying and ranking novel independent features for cardiovascular disease prediction in people with type 2 diabetes

Background

CVD prediction models do not perform well in people with diabetes. We therefore aimed to identify novel predictors for six facets of CVD, (including coronary heart disease (CHD), Ischemic stroke, heart failure (HF), and atrial fibrillation (AF)) in people with T2DM.

Methods

Analyses were conducted using the UK biobank and were stratified on history of CVD and of T2DM: 459,142 participants without diabetes or a history of CVD, 14,610 with diabetes but without CVD, and 4,432 with diabetes and a history of CVD. Replication was performed using a 20% hold-out set, ranking features on their permuted c-statistic.

Results

Out of the 600+ candidate features, we identified a subset of replicated features, ranging between 32 for CHD in people with diabetes to 184 for CVD+HF+AF in people without diabetes. Classical CVD risk factors (e.g. parental or maternal history of heart disease, or blood pressure) were relatively highly ranked for people without diabetes. The top predictors in the people with diabetes without a CVD history included: cystatin C, self-reported health satisfaction, biochemical measures of ill health (e.g. plasma albumin). For people with diabetes and a history of CVD top features were: self-reported ill health, and blood cell counts measurements (e.g. red cell distribution width). We additionally identified risk factors unique to people with diabetes, consisting of information on dietary patterns, mental health and biochemistry measures. Consideration of these novel features improved risk classification, for example per 1000 people with diabetes 133 CVD and 165 HF cases appropriately received a higher risk.

Conclusion

Through data-driven feature selection we identified a substantial number of features relevant for prediction of cardiovascular risk in people with diabetes, the majority of which related to non-classical risk factors such as mental health, general illness markers, and kidney disease.

Exploring the Role of Plasma Lipids and Statin Interventions on Multiple Sclerosis Risk and Severity: A Mendelian Randomization Study

BACKGROUND AND OBJECTIVES

There has been considerable interest in statins because of their pleiotropic effects beyond their lipid-lowering properties. Many of these pleiotropic effects are predominantly ascribed to Rho small guanosine triphosphatases (Rho GTPases) proteins. We aimed to genetically investigate the role of lipids and statin interventions on multiple sclerosis (MS) risk and severity.

METHOD

We used two-sample Mendelian randomization (MR) to investigate (1) the causal role of genetically mimic both cholesterol-dependent (through low-density lipoprotein cholesterol (LDL-C) and cholesterol biosynthesis pathway) and cholesterol-independent (through Rho GTPases) effects of statins on MS risk and MS severity, (2) the causal link between lipids (high-density lipoprotein cholesterol [HDL-C] and triglycerides [TG]) levels and MS risk and severity, and (3) the reverse causation between lipid fractions and MS risk. We used summary statistics from the Global Lipids Genetics Consortium (GLGC), eQTLGen Consortium, and the International MS Genetics Consortium (IMSGC) for lipids, expression quantitative trait loci, and MS, respectively (GLGC: n = 188,577; eQTLGen: n = 31,684; IMSGC (MS risk): n = 41,505; IMSGC (MS severity): n = 7,069).

RESULTS

The results of MR using the inverse-variance weighted method show that genetically predicted RAC2, a member of cholesterol-independent pathway (OR 0.86 [95% CI 0.78-0.95], p-value 3.80E-03), is implicated causally in reducing MS risk. We found no evidence for the causal role of LDL-C and the member of cholesterol biosynthesis pathway on MS risk. The MR results also show that lifelong higher HDL-C (OR 1.14 [95% CI 1.04-1.26], p-value 7.94E-03) increases MS risk but TG was not. Furthermore, we found no evidence for the causal role of lipids and genetically mimicked statins on MS severity. There is no evidence of reverse causation between MS risk and lipids.

DISCUSSION

Evidence from this study suggests that RAC2 is a genetic modifier of MS risk. Because RAC2 has been reported to mediate some of the pleiotropic effects of statins, we suggest that statins may reduce MS risk through a cholesterol-independent pathway (that is, RAC2-related mechanism(s)). MR analyses also support a causal effect of HDL-C on MS risk.

Unraveling the relationships between alpha- and beta-adrenergic modulation and the risk of heart failure

Background

The effects of α and ß adrenergic receptor modulation on the risk of developing heart failure (HF) remains uncertain due to a lack of randomized controlled trials. This study aimed to estimate the effects of α and ß adrenergic receptors modulation on the risk of HF and to provide proof of principle for genetic target validation studies in HF.

Methods

Genetic variants within the cis regions encoding the adrenergic receptors α1A, α2B, ß1, and ß2 associated with blood pressure in a 757,601-participant genome-wide association study (GWAS) were selected as instruments to perform a drug target Mendelian randomization study. Effects of these variants on HF risk were derived from the HERMES GWAS (542,362 controls; 40,805 HF cases).

Results

Lower α1A or ß1 activity was associated with reduced HF risk: odds ratio (OR) 0.83 (95% CI 0.74-0.93, P = 0.001) and 0.95 (95% CI 0.93-0.97, P = 8 × 10-6). Conversely, lower α2B activity was associated with increased HF risk: OR 1.09 (95% CI 1.05-1.12, P = 3 × 10-7). No evidence of an effect of lower ß2 activity on HF risk was found: OR 0.99 (95% CI 0.92-1.07, P = 0.95). Complementary analyses showed that these effects were consistent with those on left ventricular dimensions and acted independently of any potential effect on coronary artery disease.

Conclusions

This study provides genetic evidence that α1A or ß1 receptor inhibition will likely decrease HF risk, while lower α2B activity may increase this risk. Genetic variant analysis can assist with drug development for HF prevention.

DCM-PROGRESS: predicting end-stage heart failure in non-ischemic dilated cardiomyopathy patients

Aims

Patients with non-ischemic dilated cardiomyopathy (DCM) are at considerable risk for end-stage heart failure (HF), requiring close monitoring to identify early signs of disease. We aimed to develop a model to predict the 5-years risk of end-stage HF, allowing for tailored patient monitoring and management.

Methods and results

Derivation data were available from a Dutch cohort of 293 DCM patients, with external validation available from a Czech Republic cohort of 235 DCM patients. Candidate predictors spanned patient and family histories, ECG and echocardiogram measurements, and biochemistry. End-stage HF was defined as a composite of death, heart transplantation, or implantation of a ventricular assist device. Lasso and sigmoid kernel support vector machine (SVM) algorithms were trained using cross-validation. During follow-up 65 (22%) of Dutch DCM patients developed end-stage HF, with 27 (11%) cases in the Czech cohort. Out of the two considered models, the lasso model (retaining NYHA class, heart rate, systolic blood pressure, height, R-axis, and TAPSE as predictors) reached the highest discriminative performance (testing c-statistic of 0.85, 95%CI 0.58; 0.94), which was confirmed in the external validation cohort (c-statistic of 0.75, 95%CI 0.61; 0.82), compared to a c-statistic of 0.69 for the MAGGIC score. Both the MAGGIC score and the DCM-PROGRESS model slightly over-estimated the true risk, but were otherwise appropriately calibrated.

Conclusion

We developed a highly discriminative risk-prediction model for end-stage HF in DCM patients. The model was validated in two countries, suggesting the model can meaningfully improve clinical decision-making.

Triangulating evidence from observational and Mendelian randomization studies of ketone bodies for cognitive performance

BACKGROUND

Ketone bodies (KBs) are an alternative energy supply for brain functions when glucose is limited. The most abundant ketone metabolite, 3-β-hydroxybutyrate (BOHBUT), has been suggested to prevent or delay cognitive impairment, but the evidence remains unclear. We triangulated observational and Mendelian randomization (MR) studies to investigate the association and causation between KBs and cognitive function.

METHODS

In observational analyses of 5506 participants aged ≥ 45 years from the Whitehall II study, we used multiple linear regression to investigate the associations between categorized KBs and cognitive function scores. Two-sample MR was carried out using summary statistics from an in-house KBs meta-analysis between the University College London-London School of Hygiene and Tropical Medicine-Edinburgh-Bristol (UCLEB) Consortium and Kettunen et al. (N = 45,031), and publicly available summary statistics of cognitive performance and Alzheimer's disease (AD) from the Social Science Genetic Association Consortium (N = 257,841), and the International Genomics of Alzheimer's Project (N = 54,162), respectively. Both strong (P < 5 × 10-8) and suggestive (P < 1 × 10-5) sets of instrumental variables for BOHBUT were applied. Finally, we performed cis-MR on OXCT1, a well-known gene for KB catabolism.

RESULTS

BOHBUT was positively associated with general cognitive function (β = 0.26, P = 9.74 × 10-3). In MR analyses, we observed a protective effect of BOHBUT on cognitive performance (inverse variance weighted: βIVW = 7.89 × 10-2, PIVW = 1.03 × 10-2; weighted median: βW-Median = 8.65 × 10-2, PW-Median = 9.60 × 10-3) and a protective effect on AD (βIVW =  - 0.31, odds ratio: OR = 0.74, PIVW = 3.06 × 10-2). Cis-MR showed little evidence of therapeutic modulation of OXCT1 on cognitive impairment.

CONCLUSIONS

Triangulation of evidence suggests that BOHBUT has a beneficial effect on cognitive performance. Our findings raise the hypothesis that increased BOHBUT may improve general cognitive functions, delaying cognitive impairment and reducing the risk of AD.

Individualized Family Screening for Arrhythmogenic Right Ventricular Cardiomyopathy

BACKGROUND

Clinical guidelines recommend regular screening for arrhythmogenic right ventricular cardiomyopathy (ARVC) to monitor at-risk relatives, resulting in a significant burden on clinical resources. Prioritizing relatives on their probability of developing definite ARVC may provide more efficient patient care.

OBJECTIVES

The aim of this study was to determine the predictors and probability of ARVC development over time among at-risk relatives.

METHODS

A total of 136 relatives (46% men, median age 25.5 years [IQR: 15.8-44.4 years]) from the Netherlands Arrhythmogenic Cardiomyopathy Registry without definite ARVC by 2010 task force criteria were included. Phenotype was ascertained using electrocardiography, Holter monitoring, and cardiac imaging. Subjects were divided into groups with "possible ARVC" (only genetic or familial predisposition) and "borderline ARVC" (1 minor task force criterion plus genetic or familial predisposition). Cox regression was performed to determine predictors and multistate modeling to assess the probability of ARVC development. Results were replicated in an unrelated Italian cohort (57% men, median age 37.0 years [IQR: 25.4-50.4 years]).

RESULTS

At baseline, 93 subjects (68%) had possible ARVC, and 43 (32%) had borderline ARVC. Follow-up was available for 123 relatives (90%). After 8.1 years (IQR: 4.2-11.4 years), 41 (33%) had developed definite ARVC. Independent of baseline phenotype, symptomatic subjects (P = 0.014) and those 20 to 30 years of age (P = 0.002) had a higher hazard of developing definite ARVC. Furthermore, patients with borderline ARVC had a higher probability of developing definite ARVC compared with those with possible ARVC (1-year probability 13% vs 0.6%, 3-year probability 35% vs 5%; P < 0.01). External replication showed comparable results (P > 0.05).

CONCLUSIONS

Symptomatic relatives, those 20 to 30 years of age, and those with borderline ARVC have a higher probability of developing definite ARVC. These patients may benefit from more frequent follow-up, while others may be monitored less often.

Druggable proteins influencing cardiac structure and function: Implications for heart failure therapies and cancer cardiotoxicity

Dysfunction of either the right or left ventricle can lead to heart failure (HF) and subsequent morbidity and mortality. We performed a genome-wide association study (GWAS) of 16 cardiac magnetic resonance (CMR) imaging measurements of biventricular function and structure. Cis-Mendelian randomization (MR) was used to identify plasma proteins associating with CMR traits as well as with any of the following cardiac outcomes: HF, non-ischemic cardiomyopathy, dilated cardiomyopathy (DCM), atrial fibrillation, or coronary heart disease. In total, 33 plasma proteins were prioritized, including repurposing candidates for DCM and/or HF: IL18R (providing indirect evidence for IL18), I17RA, GPC5, LAMC2, PA2GA, CD33, and SLAF7. In addition, 13 of the 25 druggable proteins (52%; 95% confidence interval, 0.31 to 0.72) could be mapped to compounds with known oncological indications or side effects. These findings provide leads to facilitate drug development for cardiac disease and suggest that cardiotoxicities of several cancer treatments might represent mechanism-based adverse effects.

Biomedical consequences of elevated cholesterol-containing lipoproteins and apolipoproteins on cardiovascular and non-cardiovascular outcomes

BACKGROUND

Higher concentrations of cholesterol-containing low-density lipoprotein (LDL-C) increase the risk of cardiovascular disease (CVD). The association of LDL-C with non-CVD traits remains unclear, as are the possible independent contributions of other cholesterol-containing lipoproteins and apolipoproteins.

METHODS

Nuclear magnetic resonance spectroscopy was used to measure the cholesterol content of high density (HDL-C), very low-density (VLDL-C), intermediate-density (IDL-C), as well as low-density lipoprotein fractions, the apolipoproteins Apo-A1 and Apo-B, as well as total triglycerides (TG), remnant-cholesterol (Rem-Chol) and total cholesterol (TC). The causal effects of these exposures were assessed against 33 outcomes using univariable and multivariable Mendelian randomization (MR).

RESULTS

The majority of cholesterol containing lipoproteins and apolipoproteins affect coronary heart disease (CHD), carotid intima-media thickness, carotid plaque, C-reactive protein (CRP) and blood pressure. Multivariable MR indicated that many of these effects act independently of HDL-C, LDL-C and TG, the most frequently measured lipid fractions. Higher concentrations of TG, VLDL-C, Rem-Chol and Apo-B increased heart failure (HF) risk; often independently of LDL-C, HDL-C or TG. Finally, a subset of these exposures associated with non-CVD traits such as Alzheimer's disease (AD: HDL-C, LDL-C, IDL-C, Apo-B), type 2 diabetes (T2DM: VLDL-C, IDL-C, LDL-C), and inflammatory bowel disease (IBD: LDL-C, IDL-C).

CONCLUSIONS

The cholesterol content of a wide range of lipoprotein and apolipoproteins associate with measures of atherosclerosis, blood pressure, CRP, and CHD, with a subset affecting HF, T2DM, AD and IBD risk. Many of the observed effects appear to act independently of LDL-C, HDL-C, and TG, supporting the targeting of lipid fractions beyond LDL-C for disease prevention.

Joint Genetic Inhibition of PCSK9 and CETP and the Association With Coronary Artery Disease: A Factorial Mendelian Randomization Study

Importance

Cholesteryl ester transfer protein inhibition (CETP) has been shown to increase levels of high-density lipoprotein cholesterol (HDL-C) and reduce levels of low-density lipoprotein cholesterol (LDL-C). Current LDL-C target attainment is low, and novel phase 3 trials are underway to investigate whether CETP inhibitors result in reduction of cardiovascular disease risk in high-risk patients who may be treated with PCSK9-inhibiting agents.

Objective

To explore the associations of combined reduction of CETP and PCSK9 concentrations with risk of coronary artery disease (CAD) and other clinical and safety outcomes.

Design, Setting, and Participants

Two-sample 2 × 2 factorial Mendelian randomization study in a general population sample that includes data for UK Biobank participants of European ancestry.

Exposures

Separate genetic scores were constructed for CETP and PCSK9 plasma protein concentrations, which were combined to determine the associations of combined genetically reduced CETP and PCSK9 concentrations with disease.

Main Outcomes and Measures

Blood lipid and lipoprotein concentrations, blood pressure, CAD, age-related macular degeneration, type 2 diabetes, any stroke and ischemic stroke, Alzheimer disease, vascular dementia, heart failure, atrial fibrillation, chronic kidney disease, asthma, and multiple sclerosis.

Results

Data for 425 354 UKB participants were included; the median (IQR) age was 59 years (51-64), and 229 399 (53.9%) were female. The associations of lower CETP and lower PCSK9 concentrations with CAD are similar when scaled per 10-mg/dL reduction in LDL-C concentrations (CETP: odds ratio [OR], 0.74; 95% CI, 0.67 to 0.81; PCSK9: OR, 0.75; 95% CI, 0.71 to 0.79). Combined exposure to lower CETP and PCSK9 concentrations was associated with an additive magnitude with lipids and all outcomes, and we did not observe any nonadditive interactions, most notably for LDL-C (CETP: effect size, -1.11 mg/dL; 95% CI, -1.40 to -0.82; PCSK9: effect size, -2.13 mg/dL; 95% CI, -2.43 to -1.84; combined: effect size, -3.47 mg/dL; 95% CI, -3.76 to -3.18; P = .34 for interaction) and CAD (CETP: OR, 0.96; 95% CI, 0.94 to 1.00; PCSK9: OR, 0.94; 95% CI, 0.91 to 0.97; combined: OR, 0.90; 95% CI, 0.87 to 0.93; P = .83 for interaction). In addition, when corrected for multiple testing, lower CETP concentrations were associated with increased age-related macular degeneration (OR, 1.11; 95% CI, 1.04 to 1.19).

Conclusions and Relevance

Our results suggest that joint inhibition of CETP and PCSK9 has additive effects on lipid traits and disease risk, including a lower risk of CAD. Further research may explore whether a combination of CETP- and PCSK9-related therapeutics can benefit high-risk patients who are unable to reach treatment targets with existing options.

Therapeutic Targets for Heart Failure Identified Using Proteomics and Mendelian Randomization

BACKGROUND

Heart failure (HF) is a highly prevalent disorder for which disease mechanisms are incompletely understood. The discovery of disease-associated proteins with causal genetic evidence provides an opportunity to identify new therapeutic targets.

METHODS

We investigated the observational and causal associations of 90 cardiovascular proteins, which were measured using affinity-based proteomic assays. First, we estimated the associations of 90 cardiovascular proteins with incident heart failure by means of a fixed-effect meta-analysis of 4 population-based studies, composed of a total of 3019 participants with 732 HF events. The causal effects of HF-associated proteins were then investigated by Mendelian randomization, using cis-protein quantitative loci genetic instruments identified from genomewide association studies in more than 30 000 individuals. To improve the precision of causal estimates, we implemented an Mendelian randomization model that accounted for linkage disequilibrium between instruments and tested the robustness of causal estimates through a multiverse sensitivity analysis that included up to 120 combinations of instrument selection parameters and Mendelian randomization models per protein. The druggability of candidate proteins was surveyed, and mechanism of action and potential on-target side effects were explored with cross-trait Mendelian randomization analysis.

RESULTS

Forty-four of ninety proteins were positively associated with risk of incident HF (P<6.0×10-4). Among these, 8 proteins had evidence of a causal association with HF that was robust to multiverse sensitivity analysis: higher CSF-1 (macrophage colony-stimulating factor 1), Gal-3 (galectin-3) and KIM-1 (kidney injury molecule 1) were positively associated with risk of HF, whereas higher ADM (adrenomedullin), CHI3L1 (chitinase-3-like protein 1), CTSL1 (cathepsin L1), FGF-23 (fibroblast growth factor 23), and MMP-12 (matrix metalloproteinase-12) were protective. Therapeutics targeting ADM and Gal-3 are currently under evaluation in clinical trials, and all the remaining proteins were considered druggable, except KIM-1.

CONCLUSIONS

We identified 44 circulating proteins that were associated with incident HF, of which 8 showed evidence of a causal relationship and 7 were druggable, including adrenomedullin, which represents a particularly promising drug target. Our approach demonstrates a tractable roadmap for the triangulation of population genomic and proteomic data for the prioritization of therapeutic targets for complex human diseases.

Unravelling the Difference Between Men and Women in Post-CABG Survival

Objectives

Women have a worse prognosis after coronary artery bypass grafting (CABG) surgery compared to men. We sought to quantify to what extent this difference in post-CABG survival could be attributed to sex itself, or whether this was mediated by difference between men and women at the time of intervention. Additionally, we explored to what extent these effects were homogenous across patient subgroups.

Methods

Time to all-cause mortality was available for 102,263 CABG patients, including 20,988 (21%) women, sourced through an individual participant data meta-analysis of five cohort studies. Difference between men and women in survival duration was assessed using Kaplan-Meier estimates, and Cox's proportional hazards model.

Results

During a median follow-up of 5 years, 13,598 (13%) patients died, with women more likely to die than men: female HR 1.20 (95%CI 1.16; 1.25). We found that differences in patient characteristics at the time of CABG procedure mediated this sex effect, and accounting for these resulted in a neutral female HR 0.98 (95%CI 0.94; 1.02). Next we performed a priori defined subgroup analyses of the five most prominent mediators: age, creatinine, peripheral vascular disease, type 2 diabetes, and heart failure. We found that women without peripheral vascular disease (PVD) or women aged 70+, survived longer than men (interaction p-values 0.04 and 6 × 10-5, respectively), with an effect reversal in younger women.

Conclusion

Sex differences in post-CABG survival were readily explained by difference in patient characteristics and comorbidities. Pre-planned analyses revealed patient subgroups (aged 70+, or without PVD) of women that survived longer than men, and a subgroup of younger women with comparatively poorer survival.

Cardiovascular risk prediction in type 2 diabetes: a comparison of 22 risk scores in primary care settings

AIMS/HYPOTHESIS

We aimed to compare the performance of risk prediction scores for CVD (i.e., coronary heart disease and stroke), and a broader definition of CVD including atrial fibrillation and heart failure (CVD+), in individuals with type 2 diabetes.

METHODS

Scores were identified through a literature review and were included irrespective of the type of predicted cardiovascular outcome or the inclusion of individuals with type 2 diabetes. Performance was assessed in a contemporary, representative sample of 168,871 UK-based individuals with type 2 diabetes (age ≥18 years without pre-existing CVD+). Missing observations were addressed using multiple imputation.

RESULTS

We evaluated 22 scores: 13 derived in the general population and nine in individuals with type 2 diabetes. The Systemic Coronary Risk Evaluation (SCORE) CVD rule derived in the general population performed best for both CVD (C statistic 0.67 [95% CI 0.67, 0.67]) and CVD+ (C statistic 0.69 [95% CI 0.69, 0.70]). The C statistic of the remaining scores ranged from 0.62 to 0.67 for CVD, and from 0.64 to 0.69 for CVD+. Calibration slopes (1 indicates perfect calibration) ranged from 0.38 (95% CI 0.37, 0.39) to 0.74 (95% CI 0.72, 0.76) for CVD, and from 0.41 (95% CI 0.40, 0.42) to 0.88 (95% CI 0.86, 0.90) for CVD+. A simple recalibration process considerably improved the performance of the scores, with calibration slopes now ranging between 0.96 and 1.04 for CVD. Scores with more predictors did not outperform scores with fewer predictors: for CVD+, QRISK3 (19 variables) had a C statistic of 0.68 (95% CI 0.68, 0.69), compared with SCORE CVD (six variables) which had a C statistic of 0.69 (95% CI 0.69, 0.70). Scores specific to individuals with diabetes did not discriminate better than scores derived in the general population: the UK Prospective Diabetes Study (UKPDS) scores performed significantly worse than SCORE CVD (p value <0.001).

CONCLUSIONS/INTERPRETATION

CVD risk prediction scores could not accurately identify individuals with type 2 diabetes who experienced a CVD event in the 10 years of follow-up. All 22 evaluated models had a comparable and modest discriminative ability.

Human Genomics and Drug Development

Insights into the genetic basis of human disease are helping to address some of the key challenges in new drug development including the very high rates of failure. Here we review the recent history of an emerging, genomics-assisted approach to pharmaceutical research and development, and its relationship to Mendelian randomization (MR), a well-established analytical approach to causal inference. We demonstrate how human genomic data linked to pharmaceutically relevant phenotypes can be used for (1) drug target identification (mapping relevant drug targets to diseases), (2) drug target validation (inferring the likely effects of drug target perturbation), (3) evaluation of the effectiveness and specificity of compound-target engagement (inferring the extent to which the effects of a compound are exclusive to the target and distinguishing between on-target and off-target compound effects), and (4) the selection of end points in clinical trials (the diseases or conditions to be evaluated as trial outcomes). We show how genomics can help identify indication expansion opportunities for licensed drugs and repurposing of compounds developed to clinical phase that proved safe but ineffective for the original intended indication. We outline statistical and biological considerations in using MR for drug target validation (drug target MR) and discuss the obstacles and challenges for scaled applications of these genomics-based approaches.

Assessment of practical applicability and clinical relevance of a commonly used LDL-C polygenic score in patients with severe hypercholesterolemia

BACKGROUND AND AIMS

Low-density lipoprotein cholesterol (LDL-C) levels vary in patients with familial hypercholesterolemia (FH) and can be explained by a single deleterious genetic variant or by the aggregate effect of multiple, common small-effect variants that can be captured in a polygenic score (PS). We set out to investigate the contribution of a previously published PS to the inter-individual LDL-C variation and coronary artery disease (CAD) risk in patients with a clinical FH phenotype.

METHODS

First, in a cohort of 628 patients referred for genetic FH testing, we evaluated the distribution of a PS for LDL-C comprising 12 genetic variants. Next, we determined its association with coronary artery disease (CAD) risk using UK Biobank data.

RESULTS

The mean PS was higher in 533 FH-variant-negative patients (FH/M-) compared with 95 FH-variant carriers (1.02 vs 0.94, p < 0.001). 39% of all patients had a PS equal to the top 20% from a population-based reference cohort and these patients were less likely to carry an FH variant (OR 0.22, 95% CI 0.10-0.48) compared with patients in the lowest 20%. In UK Biobank data, the PS explained 7.4% of variance in LDL-C levels and was associated with incident CAD. Addition of PS to a prediction model using age and sex and LDL-C did not increase the c-statistic for predicting CAD risk.

CONCLUSIONS

This 12-variant PS was higher in FH/M- patients and associated with incident CAD in UK Biobank data. However, the PS did not improve predictive accuracy when added to the readily available characteristics age, sex and LDL-C, suggesting limited discriminative value for CAD.

Validation of lipid-related therapeutic targets for coronary heart disease prevention using human genetics

Drug target Mendelian randomization (MR) studies use DNA sequence variants in or near a gene encoding a drug target, that alter the target's expression or function, as a tool to anticipate the effect of drug action on the same target. Here we apply MR to prioritize drug targets for their causal relevance for coronary heart disease (CHD). The targets are further prioritized using independent replication, co-localization, protein expression profiles and data from the British National Formulary and clinicaltrials.gov. Out of the 341 drug targets identified through their association with blood lipids (HDL-C, LDL-C and triglycerides), we robustly prioritize 30 targets that might elicit beneficial effects in the prevention or treatment of CHD, including NPC1L1 and PCSK9, the targets of drugs used in CHD prevention. We discuss how this approach can be generalized to other targets, disease biomarkers and endpoints to help prioritize and validate targets during the drug development process.

Cholesteryl ester transfer protein (CETP) as a drug target for cardiovascular disease

Development of cholesteryl ester transfer protein (CETP) inhibitors for coronary heart disease (CHD) has yet to deliver licensed medicines. To distinguish compound from drug target failure, we compared evidence from clinical trials and drug target Mendelian randomization of CETP protein concentration, comparing this to Mendelian randomization of proprotein convertase subtilisin/kexin type 9 (PCSK9). We show that previous failures of CETP inhibitors are likely compound related, as illustrated by significant degrees of between-compound heterogeneity in effects on lipids, blood pressure, and clinical outcomes observed in trials. On-target CETP inhibition, assessed through Mendelian randomization, is expected to reduce the risk of CHD, heart failure, diabetes, and chronic kidney disease, while increasing the risk of age-related macular degeneration. In contrast, lower PCSK9 concentration is anticipated to decrease the risk of CHD, heart failure, atrial fibrillation, chronic kidney disease, multiple sclerosis, and stroke, while potentially increasing the risk of Alzheimer's disease and asthma. Due to distinct effects on lipoprotein metabolite profiles, joint inhibition of CETP and PCSK9 may provide added benefit. In conclusion, we provide genetic evidence that CETP is an effective target for CHD prevention but with a potential on-target adverse effect on age-related macular degeneration.

No Clinically Relevant Effect of Heart Rate Increase and Heart Rate Recovery During Exercise on Cardiovascular Disease: A Mendelian Randomization Analysis

BACKGROUND

Reduced heart rate (HR) increase (HRI), recovery (HRR), and higher resting HR are associated with cardiovascular (CV) disease, but causal inferences have not been deduced. We investigated causal effects of HRI, HRR, and resting HR on CV risk, all-cause mortality (ACM), atrial fibrillation (AF), coronary artery disease (CAD), and ischemic stroke (IS) using Mendelian Randomization.

METHODS

11 variants for HRI, 11 for HRR, and two sets of 46 and 414 variants for resting HR were obtained from four genome-wide association studies (GWASs) on UK Biobank. We performed a lookup on GWASs for CV risk and ACM in UK Biobank (N = 375,367, 5.4% cases and N = 393,165, 4.4% cases, respectively). For CAD, AF, and IS, we used publicly available summary statistics. We used a random-effects inverse-variance weighted (IVW) method and sensitivity analyses to estimate causality.

RESULTS

IVW showed a nominally significant effect of HRI on CV events (odds ratio [OR] = 1.0012, P = 4.11 × 10-2) and on CAD and AF. Regarding HRR, IVW was not significant for any outcome. The IVW method indicated statistically significant associations of resting HR with AF (OR = 0.9825, P = 9.8 × 10-6), supported by all sensitivity analyses, and a nominally significant association with IS (OR = 0.9926, P = 9.82 × 10-3).

CONCLUSION

Our findings suggest no strong evidence of an association between HRI and HRR and any outcome and confirm prior work reporting a highly significant effect of resting HR on AF. Future research is required to explore HRI and HRR associations further using more powerful predictors, when available.

PCSK9 monoclonal antibodies for the primary and secondary prevention of cardiovascular disease

BACKGROUND

Despite the availability of effective drug therapies that reduce low-density lipoprotein (LDL)-cholesterol (LDL-C), cardiovascular disease (CVD) remains an important cause of mortality and morbidity. Therefore, additional LDL-C reduction may be warranted, especially for people who are unresponsive to, or unable to take, existing LDL-C-reducing therapies. By inhibiting the proprotein convertase subtilisin/kexin type 9 (PCSK9) enzyme, monoclonal antibodies (PCSK9 inhibitors) reduce LDL-C and CVD risk.

OBJECTIVES

Primary To quantify the effects of PCSK9 inhibitors on CVD, all-cause mortality, myocardial infarction, and stroke, compared to placebo or active treatment(s) for primary and secondary prevention. Secondary To quantify the safety of PCSK9 inhibitors, with specific focus on the incidence of influenza, hypertension, type 2 diabetes, and cancer, compared to placebo or active treatment(s) for primary and secondary prevention.

SEARCH METHODS

We identified studies by systematically searching CENTRAL, MEDLINE, Embase, and Web of Science in December 2019. We also searched ClinicalTrials.gov and the International Clinical Trials Registry Platform in August 2020 and screened the reference lists of included studies. This is an update of the review first published in 2017.

SELECTION CRITERIA

All parallel-group and factorial randomised controlled trials (RCTs) with a follow-up of at least 24 weeks were eligible.

DATA COLLECTION AND ANALYSIS

Two review authors independently reviewed and extracted data. Where data were available, we calculated pooled effect estimates. We used GRADE to assess certainty of evidence and in 'Summary of findings' tables.

MAIN RESULTS

We included 24 studies with data on 60,997 participants. Eighteen trials randomised participants to alirocumab and six to evolocumab. All participants received background lipid-lowering treatment or lifestyle counselling. Six alirocumab studies used  an active treatment comparison group (the remaining used placebo), compared to three evolocumab active comparison trials. Alirocumab compared with placebo decreased the risk of CVD events, with an absolute risk difference (RD) of -2% (odds ratio (OR) 0.87, 95% confidence interval (CI) 0.80 to 0.94; 10 studies, 23,868 participants; high-certainty evidence), decreased the risk of mortality (RD -1%; OR 0.83, 95% CI 0.72 to 0.96; 12 studies, 24,797 participants; high-certainty evidence), and MI (RD -2%; OR 0.86, 95% CI 0.79 to 0.94; 9 studies, 23,352 participants; high-certainty evidence) and for any stroke (RD 0%; OR 0.73, 95% CI 0.58 to 0.91; 8 studies, 22,835 participants; high-certainty evidence). Compared to active treatment the alirocumab effects, for CVD, the RD was 1% (OR 1.37, 95% CI 0.65 to 2.87; 3 studies, 1379 participants; low-certainty evidence); for mortality, RD was -1% (OR 0.51, 95% CI 0.18 to 1.40; 5 studies, 1333 participants; low-certainty evidence); for MI, RD was 1% (OR 1.45, 95% CI 0.64 to 3.28, 5 studies, 1734 participants; low-certainty evidence); and for any stroke, RD was less than 1% (OR 0.85, 95% CI 0.13 to 5.61; 5 studies, 1734 participants; low-certainty evidence). Compared to placebo the evolocumab, for CVD, the RD was -2% (OR 0.84, 95% CI 0.78 to 0.91; 3 studies, 29,432 participants; high-certainty evidence); for mortality, RD was less than 1% (OR 1.04, 95% CI 0.91 to 1.19; 3 studies, 29,432 participants; high-certainty evidence); for MI, RD was -1% (OR 0.72, 95% CI 0.64 to 0.82; 3 studies, 29,432 participants; high-certainty evidence); and for any stroke RD was less than -1% (OR 0.79, 95% CI 0.65 to 0.94; 2 studies, 28,531 participants; high-certainty evidence).  Compared to active treatment, the evolocumab effects, for any CVD event RD was less than -1% (OR 0.66, 95% CI 0.14 to 3.04; 1 study, 218 participants; very low-certainty evidence); for all-cause mortality, the RD was less than 1% (OR 0.43, 95% CI 0.14 to 1.30; 3 studies, 5223 participants; very low-certainty evidence); and for MI, RD was less than 1% (OR 0.66, 95% CI 0.23 to 1.85; 3 studies, 5003 participants; very low-certainty evidence). There were insufficient data on any stroke.  AUTHORS' CONCLUSIONS: The evidence for the clinical endpoint effects of  evolocumab and alirocumab were graded as high. There is a strong evidence base to prescribe PCSK9 monoclonal antibodies to people who might not be eligible for other lipid-lowering drugs, or to people who cannot meet their lipid goals on more traditional therapies, which was the main patient population of the available trials.  The evidence base of PCSK9 inhibitors compared with active treatment is much weaker (low very- to low-certainty evidence) and it is unclear whether evolocumab or alirocumab might be effectively used as replacement therapies. Related, most of the available studies preferentially enrolled people with either established CVD or at a high risk already, and evidence in low- to medium-risk settings is minimal. Finally, there is very limited evidence on any potential safety issues of both evolocumab and alirocumab. While the current evidence synthesis does not reveal any adverse signals, neither does it provide evidence against such signals. This suggests careful consideration of alternative lipid lowering treatments before prescribing PCSK9 inhibitors.

Association of Factor V Leiden With Subsequent Atherothrombotic Events: A GENIUS-CHD Study of Individual Participant Data

BACKGROUND

Studies examining the role of factor V Leiden among patients at higher risk of atherothrombotic events, such as those with established coronary heart disease (CHD), are lacking. Given that coagulation is involved in the thrombus formation stage on atherosclerotic plaque rupture, we hypothesized that factor V Leiden may be a stronger risk factor for atherothrombotic events in patients with established CHD.

METHODS

We performed an individual-level meta-analysis including 25 prospective studies (18 cohorts, 3 case-cohorts, 4 randomized trials) from the GENIUS-CHD (Genetics of Subsequent Coronary Heart Disease) consortium involving patients with established CHD at baseline. Participating studies genotyped factor V Leiden status and shared risk estimates for the outcomes of interest using a centrally developed statistical code with harmonized definitions across studies. Cox proportional hazards regression models were used to obtain age- and sex-adjusted estimates. The obtained estimates were pooled using fixed-effect meta-analysis. The primary outcome was composite of myocardial infarction and CHD death. Secondary outcomes included any stroke, ischemic stroke, coronary revascularization, cardiovascular mortality, and all-cause mortality.

RESULTS

The studies included 69 681 individuals of whom 3190 (4.6%) were either heterozygous or homozygous (n=47) carriers of factor V Leiden. Median follow-up per study ranged from 1.0 to 10.6 years. A total of 20 studies with 61 147 participants and 6849 events contributed to analyses of the primary outcome. Factor V Leiden was not associated with the combined outcome of myocardial infarction and CHD death (hazard ratio, 1.03 [95% CI, 0.92-1.16]; I2=28%; P-heterogeneity=0.12). Subgroup analysis according to baseline characteristics or strata of traditional cardiovascular risk factors did not show relevant differences. Similarly, risk estimates for the secondary outcomes including stroke, coronary revascularization, cardiovascular mortality, and all-cause mortality were also close to identity.

CONCLUSIONS

Factor V Leiden was not associated with increased risk of subsequent atherothrombotic events and mortality in high-risk participants with established and treated CHD. Routine assessment of factor V Leiden status is unlikely to improve atherothrombotic events risk stratification in this population.

Genetic drug target validation using Mendelian randomisation

Mendelian randomisation (MR) analysis is an important tool to elucidate the causal relevance of environmental and biological risk factors for disease. However, causal inference is undermined if genetic variants used to instrument a risk factor also influence alternative disease-pathways (horizontal pleiotropy). Here we report how the 'no horizontal pleiotropy assumption' is strengthened when proteins are the risk factors of interest. Proteins are typically the proximal effectors of biological processes encoded in the genome. Moreover, proteins are the targets of most medicines, so MR studies of drug targets are becoming a fundamental tool in drug development. To enable such studies, we introduce a mathematical framework that contrasts MR analysis of proteins with that of risk factors located more distally in the causal chain from gene to disease. We illustrate key model decisions and introduce an analytical framework for maximising power and evaluating the robustness of analyses.

Polygenic risk scores for coronary artery disease and subsequent event risk amongst established cases

BACKGROUND

There is growing evidence that polygenic risk scores (PRSs) can identify individuals with elevated lifetime risk of coronary artery disease (CAD). Whether they can also be used to stratify the risk of subsequent events among those surviving a first CAD event remain uncertain, with possible biological differences between CAD onset and progression, and the potential for index event bias.

METHODS

Using two baseline subsamples of UK Biobank: prevalent CAD cases (N = 10 287) and individuals without CAD (N = 393 108), we evaluated associations between a CAD PRS and incident cardiovascular and fatal outcomes.

RESULTS

A 1 SD higher PRS was associated with an increased risk of incident myocardial infarction (MI) in participants without CAD (OR 1.33; 95% CI 1.29, 1.38), but the effect estimate was markedly attenuated in those with prevalent CAD (OR 1.15; 95% CI 1.06, 1.25) and heterogeneity P = 0.0012. Additionally, among prevalent CAD cases, we found an evidence of an inverse association between the CAD PRS and risk of all-cause death (OR 0.91; 95% CI 0.85, 0.98) compared with those without CAD (OR 1.01; 95% CI 0.99, 1.03) and heterogeneity P = 0.0041. A similar inverse association was found for ischaemic stroke [prevalent CAD (OR 0.78; 95% CI 0.67, 0.90); without CAD (OR 1.09; 95% CI 1.04, 1.15), heterogeneity P < 0.001].

CONCLUSIONS

Bias induced by case stratification and survival into UK Biobank may distort the associations of PRS derived from case-control studies or populations initially free of disease. Differentiating between effects of possible biases and genuine biological heterogeneity is a major challenge in disease progression research.

Lipid lowering and Alzheimer disease risk: A mendelian randomization study

Objective

To examine whether genetic variation affecting the expression or function of lipid‐lowering drug targets is associated with Alzheimer disease (AD) risk, to evaluate the potential impact of long‐term exposure to corresponding therapeutics.

Methods

We conducted Mendelian randomization analyses using variants in genes that encode the protein targets of several approved lipid‐lowering drug classes: HMGCR (encoding the target for statins), PCSK9 (encoding the target for PCSK9 inhibitors, eg, evolocumab and alirocumab), NPC1L1 (encoding the target for ezetimibe), and APOB (encoding the target of mipomersen). Variants were weighted by associations with low‐density lipoprotein cholesterol (LDL‐C) using data from lipid genetics consortia (n up to 295,826). We meta‐analyzed Mendelian randomization estimates for regional variants weighted by LDL‐C on AD risk from 2 large samples (total n = 24,718 cases, 56,685 controls).

Results

Models for HMGCR, APOB, and NPC1L1 did not suggest that the use of related lipid‐lowering drug classes would affect AD risk. In contrast, genetically instrumented exposure to PCSK9 inhibitors was predicted to increase AD risk in both of the AD samples (combined odds ratio per standard deviation lower LDL‐C inducible by the drug target = 1.45, 95% confidence interval = 1.23–1.69). This risk increase was opposite to, although more modest than, the degree of protection from coronary artery disease predicted by these same methods for PCSK9 inhibition.

Interpretation

We did not identify genetic support for the repurposing of statins, ezetimibe, or mipomersen for AD prevention. Notwithstanding caveats to this genetic evidence, pharmacovigilance for AD risk among users of PCSK9 inhibitors may be warranted. ANN NEUROL 2020;87:30–39

Association between 8 P-glycoprotein (MDR1/ABCB1) gene polymorphisms and antipsychotic drug-induced hyperprolactinaemia

Introduction

Hyperprolactinaemia, a common adverse effect of antipsychotic drugs, is primarily linked to blockade of dopamine D2 receptors in the pituitary gland. Certain antipsychotic drugs, such as, for example risperidone and paliperidone, are more likely to induce hyperprolactinaemia compared to others. This effect is probably caused by a relatively high blood/brain concentration ratio, a consequence of being a substrate of P‐glycoprotein. Genetic variants of P‐glycoprotein with changed functional activity might influence the potential of risperidone and paliperidone to cause hyperprolactinaemia as the altered blood/brain concentration ratio would lead to a reduced therapeutic drug level within essential brain areas making dose adaptations necessary. This increases exposure of dopamine D2 receptors within the pituitary gland.

Aims

To investigate possible associations between MDR1/ABCB1 gene polymorphisms and antipsychotic drug‐induced hyperprolactinaemia in Russian patients with schizophrenia and to determine possible differences between risperidone/paliperidone and other antipsychotics.

Methods

In total, 446 patients with schizophrenia were included from 3 psychiatric hospitals in Siberia. Blood samples were obtained in a cross‐sectional study design for DNA extraction and prolactin measurement. Associations between hyperprolactinaemia and 8 MDR1/ABCB1 gene‐polymorphisms were assessed using logistic regression analysis accounting for covariates. The analysis was repeated in a patient subgroup using risperidone or paliperidone.

Results

We did not observe an association between any of the 8 single nucleotide polymorphisms and the prevalence of antipsychotic‐induced hyperprolactinaemia in the total patient population. However, in the risperidone/paliperidone subgroup, the single nucleotide polymorphism rs2032582 (G2677T) was found to be negatively associated with risperidone/paliperidone‐induced hyperprolactinaemia.

Conclusion

This study revealed a significant association between the ABCB1 gene polymorphism rs2032582 (G2677T) and risperidone/paliperidone‐induced hyperprolactinaemia.

The median and the mode as robust meta-analysis estimators in the presence of small-study effects and outliers

Meta‐analyses based on systematic literature reviews are commonly used to obtain a quantitative summary of the available evidence on a given topic. However, the reliability of any meta‐analysis is constrained by that of its constituent studies. One major limitation is the possibility of small‐study effects, when estimates from smaller and larger studies differ systematically. Small‐study effects may result from reporting biases (ie, publication bias), from inadequacies of the included studies that are related to study size, or from reasons unrelated to bias. We propose two estimators based on the median and mode to increase the reliability of findings in a meta‐analysis by mitigating the influence of small‐study effects. By re‐examining data from published meta‐analyses and by conducting a simulation study, we show that these estimators offer robustness to a range of plausible bias mechanisms, without making explicit modelling assumptions. They are also robust to outlying studies without explicitly removing such studies from the analysis. When meta‐analyses are suspected to be at risk of bias because of small‐study effects, we recommend reporting the mean, median and modal pooled estimates.

Association Between BDNF Gene Variant Rs6265 and the Severity of Depression in Antidepressant Treatment-Free Depressed Patients

BACKGROUND

Brain-derived neurotrophic factor (BDNF) plays an important role in neuronal plasticity, and its dysregulation has been associated with the pathogenesis of mood and anxiety disorders. Prolactin (PRL) is a pituitary hormone which is also produced as a cytokine by immune cells and could be a neurotrophic factor regulating the functional activity of stress-related mechanisms.

AIM

To investigate the possible relationship between depressive state and BDNF and PRL genotypes or levels with special reference to severity of depression.

METHODS

Participants of 18-70 years with a clinical diagnosis of depressive disorder of at least moderate severity were included. These patients had not been treated with antidepressant drugs before admission to hospital during the preceding period of the last 6 months, and 54.5% had never been treated with antidepressant drugs during their entire life. The DNA was genotyped for rs1341239 within the prolactin and for rs6265, rs7124442, and rs11030104 within the BDNF gene. Rs11030104 violated the Hardy-Weinberg equilibrium distribution and was excluded from further analyses. BDNF and prolactin concentration was measured in serum by MAGPIX multiplex analyzer (Luminex, USA) using MILLIPLEX^® MAP kit (Merck, Germany). Genetic associations were determined by sequentially regressing prolactin, BDNF, 17-items Hamilton's Depression (HAMD-17) and Clinical Global Impression scale, Severity (CGI-S) ratings, and depression (absent/present) on the available SNPs. Genetic associations were evaluated assuming an additive model.

RESULTS

A total of 186 depressed patients (of which 169 were women) and 94 healthy controls (67 women) were genotyped. After excluding subjects without genetic information on all three study SNPs, 217 remained of whom 138 suffered from depression. Within depressed patients we observed an association of rs6265 with HAMD-17: mean difference (MD) 2.33 (95%CI 0.49; 4.16; p = 0.014) and CGI-S: MD 0.38 (95%CI 0.09; 0.66; p = 0.011). No significant association was observed between the prolactin SNP rs1341239 and prolactin levels. Similarly the mean differences of BDNF SNPs did not show an association with BDNF: rs6265 -0.042 ln(pg/ml) (95%CI -0.198; 0.113), and rs7124442 0.006 ln(pg/ml) (95%CI -0.117; 0.130). No other association reached statistical significance.

CONCLUSION

We observed a significant association between BDNF gene variant rs6265 and the severity of depression in newly admitted, antidepressant treatment-free, depressed patients. Actual PRL and BDNF levels were not elevated sufficiently in depressed patients to reach statistical significance and were not associated with the studied genotypes.

Phenome-wide association analysis of LDL-cholesterol lowering genetic variants in PCSK9

BACKGROUND

We characterised the phenotypic consequence of genetic variation at the PCSK9 locus and compared findings with recent trials of pharmacological inhibitors of PCSK9.

METHODS

Published and individual participant level data (300,000+ participants) were combined to construct a weighted PCSK9 gene-centric score (GS). Seventeen randomized placebo controlled PCSK9 inhibitor trials were included, providing data on 79,578 participants. Results were scaled to a one mmol/L lower LDL-C concentration.

RESULTS

The PCSK9 GS (comprising 4 SNPs) associations with plasma lipid and apolipoprotein levels were consistent in direction with treatment effects. The GS odds ratio (OR) for myocardial infarction (MI) was 0.53 (95% CI 0.42; 0.68), compared to a PCSK9 inhibitor effect of 0.90 (95% CI 0.86; 0.93). For ischemic stroke ORs were 0.84 (95% CI 0.57; 1.22) for the GS, compared to 0.85 (95% CI 0.78; 0.93) in the drug trials. ORs with type 2 diabetes mellitus (T2DM) were 1.29 (95% CI 1.11; 1.50) for the GS, as compared to 1.00 (95% CI 0.96; 1.04) for incident T2DM in PCSK9 inhibitor trials. No genetic associations were observed for cancer, heart failure, atrial fibrillation, chronic obstructive pulmonary disease, or Alzheimer's disease - outcomes for which large-scale trial data were unavailable.

CONCLUSIONS

Genetic variation at the PCSK9 locus recapitulates the effects of therapeutic inhibition of PCSK9 on major blood lipid fractions and MI. While indicating an increased risk of T2DM, no other possible safety concerns were shown; although precision was moderate.

Mendelian randomization with Egger pleiotropy correction and weakly informative Bayesian priors

Background

The MR-Egger (MRE) estimator has been proposed to correct for directional pleiotropic effects of genetic instruments in an instrumental variable (IV) analysis. The power of this method is considerably lower than that of conventional estimators, limiting its applicability. Here we propose a novel Bayesian implementation of the MR-Egger estimator (BMRE) and explore the utility of applying weakly informative priors on the intercept term (the pleiotropy estimate) to increase power of the IV (slope) estimate.

Methods

This was a simulation study to compare the performance of different IV estimators. Scenarios differed in the presence of a causal effect, the presence of pleiotropy, the proportion of pleiotropic instruments and degree of 'Instrument Strength Independent of Direct Effect' (InSIDE) assumption violation. Based on empirical plasma urate data, we present an approach to elucidate a prior distribution for the amount of pleiotropy.

Results

A weakly informative prior on the intercept term increased power of the slope estimate while maintaining type 1 error rates close to the nominal value of 0.05. Under the InSIDE assumption, performance was unaffected by the presence or absence of pleiotropy. Violation of the InSIDE assumption biased all estimators, affecting the BMRE more than the MRE method.

Conclusions

Depending on the prior distribution, the BMRE estimator has more power at the cost of an increased susceptibility to InSIDE assumption violations. As such the BMRE method is a compromise between the MRE and conventional IV estimators, and may be an especially useful approach to account for observed pleiotropy.

Subsequent Event Risk in Individuals With Established Coronary Heart Disease

Supplemental Digital Content is available in the text.

Background:

The Genetics of Subsequent Coronary Heart Disease (GENIUS-CHD) consortium was established to facilitate discovery and validation of genetic variants and biomarkers for risk of subsequent CHD events, in individuals with established CHD.

Methods:

The consortium currently includes 57 studies from 18 countries, recruiting 185 614 participants with either acute coronary syndrome, stable CHD, or a mixture of both at baseline. All studies collected biological samples and followed-up study participants prospectively for subsequent events.

Results:

Enrollment into the individual studies took place between 1985 to present day with a duration of follow-up ranging from 9 months to 15 years. Within each study, participants with CHD are predominantly of self-reported European descent (38%–100%), mostly male (44%–91%) with mean ages at recruitment ranging from 40 to 75 years. Initial feasibility analyses, using a federated analysis approach, yielded expected associations between age (hazard ratio, 1.15; 95% CI, 1.14–1.16) per 5-year increase, male sex (hazard ratio, 1.17; 95% CI, 1.13–1.21) and smoking (hazard ratio, 1.43; 95% CI, 1.35–1.51) with risk of subsequent CHD death or myocardial infarction and differing associations with other individual and composite cardiovascular endpoints.

Conclusions:

GENIUS-CHD is a global collaboration seeking to elucidate genetic and nongenetic determinants of subsequent event risk in individuals with established CHD, to improve residual risk prediction and identify novel drug targets for secondary prevention. Initial analyses demonstrate the feasibility and reliability of a federated analysis approach. The consortium now plans to initiate and test novel hypotheses as well as supporting replication and validation analyses for other investigators.

Association of Chromosome 9p21 With Subsequent Coronary Heart Disease Events

Supplemental Digital Content is available in the text.

Background:

Genetic variation at chromosome 9p21 is a recognized risk factor for coronary heart disease (CHD). However, its effect on disease progression and subsequent events is unclear, raising questions about its value for stratification of residual risk.

Methods:

A variant at chromosome 9p21 (rs1333049) was tested for association with subsequent events during follow-up in 103 357 Europeans with established CHD at baseline from the GENIUS-CHD (Genetics of Subsequent Coronary Heart Disease) Consortium (73.1% male, mean age 62.9 years). The primary outcome, subsequent CHD death or myocardial infarction (CHD death/myocardial infarction), occurred in 13 040 of the 93 115 participants with available outcome data. Effect estimates were compared with case/control risk obtained from the CARDIoGRAMplusC4D consortium (Coronary Artery Disease Genome-wide Replication and Meta-analysis [CARDIoGRAM] plus The Coronary Artery Disease [C4D] Genetics) including 47 222 CHD cases and 122 264 controls free of CHD.

Results:

Meta-analyses revealed no significant association between chromosome 9p21 and the primary outcome of CHD death/myocardial infarction among those with established CHD at baseline (GENIUS-CHD odds ratio, 1.02; 95% CI, 0.99–1.05). This contrasted with a strong association in CARDIoGRAMPlusC4D odds ratio 1.20; 95% CI, 1.18–1.22; P for interaction <0.001 compared with the GENIUS-CHD estimate. Similarly, no clear associations were identified for additional subsequent outcomes, including all-cause death, although we found a modest positive association between chromosome 9p21 and subsequent revascularization (odds ratio, 1.07; 95% CI, 1.04–1.09).

Conclusions:

In contrast to studies comparing individuals with CHD to disease-free controls, we found no clear association between genetic variation at chromosome 9p21 and risk of subsequent acute CHD events when all individuals had CHD at baseline. However, the association with subsequent revascularization may support the postulated mechanism of chromosome 9p21 for promoting atheroma development.

An electronic health records cohort study on heart failure following myocardial infarction in England: incidence and predictors

OBJECTIVES

To investigate the incidence and determinants of heart failure (HF) following a myocardial infarction (MI) in a contemporary cohort of patients with MI using routinely collected primary and hospital care electronic health records (EHRs).

METHODS

Data were used from the CALIBER programme, linking EHRs in England from primary care, hospital admissions, an MI registry and mortality data. Subjects were eligible if they were 18 years or older, did not have a history of HF and survived a first MI. Factors associated with time to HF were examined using Cox proportional hazard models.

RESULTS

Of the 24 479 patients with MI, 5775 (23.6%) developed HF during a median follow-up of 3.7 years (incidence rate per 1000 person-years: 63.8, 95% CI 62.2 to 65.5). Baseline characteristics significantly associated with developing HF were: atrial fibrillation (HR 1.62, 95% CI 1.51 to 1.75), age (per 10 years increase: 1.45, 1.41 to 1.49), diabetes (1.45, 1.35 to 1.56), peripheral arterial disease (1.38, 1.26 to 1.51), chronic obstructive pulmonary disease (1.28, 1.17 to 1.40), greater socioeconomic deprivation (5th vs 1st quintile: 1.27, 1.13 to 1.41), ST-segment elevation MI at presentation (1.19, 1.11 to 1.27) and hypertension (1.16, 1.09 to 1.23). Results were robust to various sensitivity analyses such as competing risk analysis and multiple imputation.

CONCLUSION

In England, one in four survivors of a first MI develop HF within 4 years. This contemporary study demonstrates that patients with MI are at considerable risk of HF. Baseline patient characteristics associated with time until HF were identified, which may be used to target preventive strategies.

Comparison of variance estimators for meta-analysis of instrumental variable estimates

Background: Mendelian randomization studies perform instrumental variable (IV) analysis using genetic IVs. Results of individual Mendelian randomization studies can be pooled through meta-analysis. We explored how different variance estimators influence the meta-analysed IV estimate.

Methods: Two versions of the delta method (IV before or after pooling), four bootstrap estimators, a jack-knife estimator and a heteroscedasticity-consistent (HC) variance estimator were compared using simulation. Two types of meta-analyses were compared, a two-stage meta-analysis pooling results, and a one-stage meta-analysis pooling datasets.

Results: Using a two-stage meta-analysis, coverage of the point estimate using bootstrapped estimators deviated from nominal levels at weak instrument settings and/or outcome probabilities ≤ 0.10. The jack-knife estimator was the least biased resampling method, the HC estimator often failed at outcome probabilities ≤ 0.50 and overall the delta method estimators were the least biased. In the presence of between-study heterogeneity, the delta method before meta-analysis performed best. Using a one-stage meta-analysis all methods performed equally well and better than two-stage meta-analysis of greater or equal size.

Conclusions: In the presence of between-study heterogeneity, two-stage meta-analyses should preferentially use the delta method before meta-analysis. Weak instrument bias can be reduced by performing a one-stage meta-analysis.

Linear regression and the normality assumption

OBJECTIVES

Researchers often perform arbitrary outcome transformations to fulfill the normality assumption of a linear regression model. This commentary explains and illustrates that in large data settings, such transformations are often unnecessary, and worse may bias model estimates.

STUDY DESIGN AND SETTING

Linear regression assumptions are illustrated using simulated data and an empirical example on the relation between time since type 2 diabetes diagnosis and glycated hemoglobin levels. Simulation results were evaluated on coverage; i.e., the number of times the 95% confidence interval included the true slope coefficient.

RESULTS

Although outcome transformations bias point estimates, violations of the normality assumption in linear regression analyses do not. The normality assumption is necessary to unbiasedly estimate standard errors, and hence confidence intervals and P-values. However, in large sample sizes (e.g., where the number of observations per variable is >10) violations of this normality assumption often do not noticeably impact results. Contrary to this, assumptions on, the parametric model, absence of extreme observations, homoscedasticity, and independency of the errors, remain influential even in large sample size settings.

CONCLUSION

Given that modern healthcare research typically includes thousands of subjects focusing on the normality assumption is often unnecessary, does not guarantee valid results, and worse may bias estimates due to the practice of outcome transformations.

Impact of Selection Bias on Estimation of Subsequent Event Risk

BACKGROUND

Studies of recurrent or subsequent disease events may be susceptible to bias caused by selection of subjects who both experience and survive the primary indexing event. Currently, the magnitude of any selection bias, particularly for subsequent time-to-event analysis in genetic association studies, is unknown.

METHODS AND RESULTS

We used empirically inspired simulation studies to explore the impact of selection bias on the marginal hazard ratio for risk of subsequent events among those with established coronary heart disease. The extent of selection bias was determined by the magnitudes of genetic and nongenetic effects on the indexing (first) coronary heart disease event. Unless the genetic hazard ratio was unrealistically large (>1.6 per allele) and assuming the sum of all nongenetic hazard ratios was <10, bias was usually <10% (downward toward the null). Despite the low bias, the probability that a confidence interval included the true effect decreased (undercoverage) with increasing sample size because of increasing precision. Importantly, false-positive rates were not affected by selection bias.

CONCLUSIONS

In most empirical settings, selection bias is expected to have a limited impact on genetic effect estimates of subsequent event risk. Nevertheless, because of undercoverage increasing with sample size, most confidence intervals will be over precise (not wide enough). When there is no effect modification by history of coronary heart disease, the false-positive rates of association tests will be close to nominal.

Relations between lipoprotein(a) concentrations, LPA genetic variants, and the risk of mortality in patients with established coronary heart disease: a molecular and genetic association study

BACKGROUND

Lipoprotein(a) concentrations in plasma are associated with cardiovascular risk in the general population. Whether lipoprotein(a) concentrations or LPA genetic variants predict long-term mortality in patients with established coronary heart disease remains less clear.

METHODS

We obtained data from 3313 patients with established coronary heart disease in the Ludwigshafen Risk and Cardiovascular Health (LURIC) study. We tested associations of tertiles of lipoprotein(a) concentration in plasma and two LPA single-nucleotide polymorphisms ([SNPs] rs10455872 and rs3798220) with all-cause mortality and cardiovascular mortality by Cox regression analysis and with severity of disease by generalised linear modelling, with and without adjustment for age, sex, diabetes diagnosis, systolic blood pressure, BMI, smoking status, estimated glomerular filtration rate, LDL-cholesterol concentration, and use of lipid-lowering therapy. Results for plasma lipoprotein(a) concentrations were validated in five independent studies involving 10 195 patients with established coronary heart disease. Results for genetic associations were replicated through large-scale collaborative analysis in the GENIUS-CHD consortium, comprising 106 353 patients with established coronary heart disease and 19 332 deaths in 22 studies or cohorts.

FINDINGS

The median follow-up was 9·9 years. Increased severity of coronary heart disease was associated with lipoprotein(a) concentrations in plasma in the highest tertile (adjusted hazard radio [HR] 1·44, 95% CI 1·14-1·83) and the presence of either LPA SNP (1·88, 1·40-2·53). No associations were found in LURIC with all-cause mortality (highest tertile of lipoprotein(a) concentration in plasma 0·95, 0·81-1·11 and either LPA SNP 1·10, 0·92-1·31) or cardiovascular mortality (0·99, 0·81-1·2 and 1·13, 0·90-1·40, respectively) or in the validation studies.

INTERPRETATION

In patients with prevalent coronary heart disease, lipoprotein(a) concentrations and genetic variants showed no associations with mortality. We conclude that these variables are not useful risk factors to measure to predict progression to death after coronary heart disease is established.

FUNDING

Seventh Framework Programme for Research and Technical Development (AtheroRemo and RiskyCAD), INTERREG IV Oberrhein Programme, Deutsche Nierenstiftung, Else-Kroener Fresenius Foundation, Deutsche Stiftung für Herzforschung, Deutsche Forschungsgemeinschaft, Saarland University, German Federal Ministry of Education and Research, Willy Robert Pitzer Foundation, and Waldburg-Zeil Clinics Isny.

PCSK9 monoclonal antibodies for the primary and secondary prevention of cardiovascular disease

BACKGROUND

Despite the availability of effective drug therapies that reduce low-density lipoprotein (LDL)-cholesterol (LDL-C), cardiovascular disease (CVD) remains an important cause of mortality and morbidity. Therefore, additional LDL-C reduction may be warranted, especially for patients who are unresponsive to, or unable to take, existing LDL-C-reducing therapies. By inhibiting the proprotein convertase subtilisin/kexin type 9 (PCSK9) enzyme, monoclonal antibodies (PCSK9 inhibitors) may further reduce LDL-C, potentially reducing CVD risk as well.

OBJECTIVES

Primary To quantify short-term (24 weeks), medium-term (one year), and long-term (five years) effects of PCSK9 inhibitors on lipid parameters and on the incidence of CVD. Secondary To quantify the safety of PCSK9 inhibitors, with specific focus on the incidence of type 2 diabetes, cognitive function, and cancer. Additionally, to determine if specific patient subgroups were more or less likely to benefit from the use of PCSK9 inhibitors.

SEARCH METHODS

We identified studies by systematically searching the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, and Web of Science. We also searched Clinicaltrials.gov and the International Clinical Trials Registry Platform and screened the reference lists of included studies. We identified the studies included in this review through electronic literature searches conducted up to May 2016, and added three large trials published in March 2017.

SELECTION CRITERIA

All parallel-group and factorial randomised controlled trials (RCTs) with a follow-up time of at least 24 weeks were eligible.

DATA COLLECTION AND ANALYSIS

Two review authors independently reviewed and extracted data. When data were available, we calculated pooled effect estimates.

MAIN RESULTS

We included 20 studies with data on 67,237 participants (median age 61 years; range 52 to 64 years). Twelve trials randomised participants to alirocumab, three trials to bococizumab, one to RG7652, and four to evolocumab. Owing to the small number of trials using agents other than alirocumab, we did not differentiate between types of PCSK9 inhibitors used. We compared PCSK9 inhibitors with placebo (thirteen RCTs), ezetimibe (two RCTs) or ezetimibe and statins (five RCTs).Compared with placebo, PCSK9 inhibitors decreased LDL-C by 53.86% (95% confidence interval (CI) 58.64 to 49.08; eight studies; 4782 participants; GRADE: moderate) at 24 weeks; compared with ezetimibe, PCSK9 inhibitors decreased LDL-C by 30.20% (95% CI 34.18 to 26.23; two studies; 823 participants; GRADE: moderate), and compared with ezetimibe and statins, PCSK9 inhibitors decreased LDL-C by 39.20% (95% CI 56.15 to 22.26; five studies; 5376 participants; GRADE: moderate).Compared with placebo, PCSK9 inhibitors decreased the risk of CVD events, with a risk difference (RD) of 0.91% (odds ratio (OR) of 0.86, 95% CI 0.80 to 0.92; eight studies; 59,294 participants; GRADE: moderate). Compared with ezetimibe and statins, PCSK9 inhibitors appeared to have a stronger protective effect on CVD risk, although with considerable uncertainty (RD 1.06%, OR 0.45, 95% CI 0.27 to 0.75; three studies; 4770 participants; GRADE: very low). No data were available for the ezetimibe only comparison. Compared with placebo, PCSK9 probably had little or no effect on mortality (RD 0.03%, OR 1.02, 95% CI 0.91 to 1.14; 12 studies; 60,684 participants; GRADE: moderate). Compared with placebo, PCSK9 inhibitors increased the risk of any adverse events (RD 1.54%, OR 1.08, 95% CI 1.04 to 1.12; 13 studies; 54,204 participants; GRADE: low). Similar effects were observed for the comparison of ezetimibe and statins: RD 3.70%, OR 1.18, 95% CI 1.05 to 1.34; four studies; 5376 participants; GRADE: low. Clinical event data were unavailable for the ezetimibe only comparison.

AUTHORS' CONCLUSIONS

Over short-term to medium-term follow-up, PCSK9 inhibitors reduced LDL-C. Studies with medium-term follow-up time (longest median follow-up recorded was 26 months) reported that PCSK9 inhibitors (compared with placebo) decreased CVD risk but may have increased the risk of any adverse events (driven by SPIRE-1 and -2 trials). Available evidence suggests that PCSK9 inhibitor use probably leads to little or no difference in mortality. Evidence on relative efficacy and safety when PCSK9 inhibitors were compared with active treatments was of low to very low quality (GRADE); follow-up times were short and events were few. Large trials with longer follow-up are needed to evaluate PCSK9 inhibitors versus active treatments as well as placebo. Owing to the predominant inclusion of high-risk patients in these studies, applicability of results to primary prevention is limited. Finally, estimated risk differences indicate that PCSK9 inhibitors only modestly change absolute risks (often to less than 1%).

PCSK9 genetic variants and risk of type 2 diabetes: a mendelian randomisation study

BACKGROUND

Statin treatment and variants in the gene encoding HMG-CoA reductase are associated with reductions in both the concentration of LDL cholesterol and the risk of coronary heart disease, but also with modest hyperglycaemia, increased bodyweight, and modestly increased risk of type 2 diabetes, which in no way offsets their substantial benefits. We sought to investigate the associations of LDL cholesterol-lowering PCSK9 variants with type 2 diabetes and related biomarkers to gauge the likely effects of PCSK9 inhibitors on diabetes risk.

METHODS

In this mendelian randomisation study, we used data from cohort studies, randomised controlled trials, case control studies, and genetic consortia to estimate associations of PCSK9 genetic variants with LDL cholesterol, fasting blood glucose, HbA1c, fasting insulin, bodyweight, waist-to-hip ratio, BMI, and risk of type 2 diabetes, using a standardised analysis plan, meta-analyses, and weighted gene-centric scores.

FINDINGS

Data were available for more than 550 000 individuals and 51 623 cases of type 2 diabetes. Combined analyses of four independent PCSK9 variants (rs11583680, rs11591147, rs2479409, and rs11206510) scaled to 1 mmol/L lower LDL cholesterol showed associations with increased fasting glucose (0·09 mmol/L, 95% CI 0·02 to 0·15), bodyweight (1·03 kg, 0·24 to 1·82), waist-to-hip ratio (0·006, 0·003 to 0·010), and an odds ratio for type diabetes of 1·29 (1·11 to 1·50). Based on the collected data, we did not identify associations with HbA1c (0·03%, -0·01 to 0·08), fasting insulin (0·00%, -0·06 to 0·07), and BMI (0·11 kg/m2, -0·09 to 0·30).

INTERPRETATION

PCSK9 variants associated with lower LDL cholesterol were also associated with circulating higher fasting glucose concentration, bodyweight, and waist-to-hip ratio, and an increased risk of type 2 diabetes. In trials of PCSK9 inhibitor drugs, investigators should carefully assess these safety outcomes and quantify the risks and benefits of PCSK9 inhibitor treatment, as was previously done for statins.

FUNDING

British Heart Foundation, and University College London Hospitals NHS Foundation Trust (UCLH) National Institute for Health Research (NIHR) Biomedical Research Centre.

Adjusting for bias in unblinded randomized controlled trials

It may not always be possible to blind participants of a randomized controlled trial for treatment allocation. As a result, estimators of the actual treatment effect may be biased. In this paper, we will extend a novel method, originally introduced in genetic research, for instrumental variable meta-analysis, adjusting for bias due to unblinding of trial participants. Using simulation studies, this novel method, "Egger Correction for non-Adherence", is introduced and compared to the performance of the "intention-to-treat," "as-treated," and conventional "instrumental variable" estimators. Scenarios considered (time-varying) non-adherence, confounding, and between-study heterogeneity. The effect of treatment on a binary endpoint was quantified by means of a risk difference. In all scenarios with unblinded treatment allocation, the Egger Correction for non-Adherence method was the least biased estimator. However, unless the variation in adherence was relatively large, precision was lacking, and power did not surpass 0.50. As a comparison, in a meta-analysis of blinded randomized controlled trials, power of the conventional IV estimator was 1.00 versus at most 0.14 for the Egger Correction for non-Adherence estimator. Due to this lack of precision and power, we suggest to use this method mainly as a sensitivity analysis.

Chemotherapy effectiveness and mortality prediction in surgically treated osteosarcoma dogs: A validation study

Canine osteosarcoma is the most common bone cancer, and an important cause of mortality and morbidity, in large purebred dogs. Previously we constructed two multivariable models to predict a dog's 5-month or 1-year mortality risk after surgical treatment for osteosarcoma. According to the 5-month model, dogs with a relatively low risk of 5-month mortality benefited most from additional chemotherapy treatment. In the present study, we externally validated these results using an independent cohort study of 794 dogs. External performance of our prediction models showed some disagreement between observed and predicted risk, mean difference: -0.11 (95% confidence interval [95% CI]-0.29; 0.08) for 5-month risk and 0.25 (95%CI 0.10; 0.40) for 1-year mortality risk. After updating the intercept, agreement improved: -0.0004 (95%CI-0.16; 0.16) and -0.002 (95%CI-0.15; 0.15). The chemotherapy by predicted mortality risk interaction (P-value=0.01) showed that the chemotherapy compared to no chemotherapy effectiveness was modified by 5-month mortality risk: dogs with a relatively lower risk of mortality benefited most from additional chemotherapy. Chemotherapy effectiveness on 1-year mortality was not significantly modified by predicted risk (P-value=0.28). In conclusion, this external validation study confirmed that our multivariable risk prediction models can predict a patient's mortality risk and that dogs with a relatively lower risk of 5-month mortality seem to benefit most from chemotherapy.

Exploring interaction effects in small samples increases rates of false-positive and false-negative findings: results from a systematic review and simulation study

OBJECTIVE

To give a comprehensive comparison of the performance of commonly applied interaction tests.

METHODS

A literature review and simulation study was performed evaluating interaction tests on the odds ratio (OR) or the risk difference (RD) scales: Cochran Q (Q), Breslow-Day (BD), Tarone, unconditional score, likelihood ratio (LR), Wald, and relative excess risk due to interaction (RERI)-based tests.

RESULTS

Review results agreed with results from our simulation study, which showed that on the OR scale, in small sample sizes (eg, number of subjects ≤ 250) the type 1 error rates of the LR test was 0.10; the BD and Tarone tests showed results around 0.05. On the RD scale, the LR and RERI tests had error rates around 0.05. On both scales, tests did not differ regarding power. When exposure prevented the outcome RERI-based tests were relatively underpowered (eg, N = 100; RERI power = 5% vs. Wald power = 18%). With increasing sample size, difference decreased.

CONCLUSION

In small samples, interaction tests differed. On the OR scale, the Tarone and BD tests are recommended. On the RD scale, the LR and RERI-based tests performed best. However, RERI-based tests are underpowered compared with other tests, when exposure prevents the outcome, and sample size is limited.