Cardiovascular risk estimation in 2012: lessons learned and applicability to the HIV population

Probability estimation with machine learning methods for dichotomous and multicategory outcome: theory

Probability estimation for binary and multicategory outcome using logistic and multinomial logistic regression has a long-standing tradition in biostatistics. However, biases may occur if the model is misspecified. In contrast, outcome probabilities for individuals can be estimated consistently with machine learning approaches, including k-nearest neighbors (k-NN), bagged nearest neighbors (b-NN), random forests (RF), and support vector machines (SVM). Because machine learning methods are rarely used by applied biostatisticians, the primary goal of this paper is to explain the concept of probability estimation with these methods and to summarize recent theoretical findings. Probability estimation in k-NN, b-NN, and RF can be embedded into the class of nonparametric regression learning machines; therefore, we start with the construction of nonparametric regression estimates and review results on consistency and rates of convergence. In SVMs, outcome probabilities for individuals are estimated consistently by repeatedly solving classification problems. For SVMs we review classification problem and then dichotomous probability estimation. Next we extend the algorithms for estimating probabilities using k-NN, b-NN, and RF to multicategory outcomes and discuss approaches for the multicategory probability estimation problem using SVM. In simulation studies for dichotomous and multicategory dependent variables we demonstrate the general validity of the machine learning methods and compare it with logistic regression. However, each method fails in at least one simulation scenario. We conclude with a discussion of the failures and give recommendations for selecting and tuning the methods. Applications to real data and example code are provided in a companion article (doi:10.1002/bimj.201300077).

HIV infection and cardiovascular disease

AIMS

With the success of antiretroviral therapy (ART), non-human immunodeficiency virus (HIV)-related comorbidities like cardiovascular diseases (CVDs) are of increasing concern. We describe important recent research developments on the epidemiology of CVD in HIV infection, ART-related metabolic changes, and cardioprotective anti-inflammatory mechanisms, and summarize management strategies for CVD risk reduction.

METHODS AND RESULTS

We systematically identified and analysed systematic reviews and most cited literature published in the last 3 years and supplemented findings with selected evidence based on clinical expertise. Among HIV-infected individuals, the prevalence of CVD risk factors and the risk for CVD is higher compared with HIV negatives. Antiretroviral drugs may induce dyslipidaemia, reduce insulin sensitivity, and promote body fat redistribution that additionally contributes to CVD risk. Some antiretroviral drugs may increase risk for CVD events, but the absolute risk increase is moderate and has to be put into perspective with the massive HIV-related benefits. Sustained HIV suppression reduces systemic inflammatory markers and is associated with a moderate reduction in CVD events. Regular CVD risk assessment and counselling to stop smoking must be regularly done in all HIV-infected individuals. Statins are effective for the treatment of dyslipidaemia in HIV infection, but drug interactions with ART need to be considered.

CONCLUSION

Human immunodeficiency virus-infected individuals are at increased risk for CVD. Timely initiation of ART with consequent viral suppression is likely to reduce CVD events and to offset potential side effects from ART-induced metabolic changes. Reduction in smoking in HIV-infected individuals is a public health priority.

Mind the gap: difference between Framingham heart age and real age increases with age in HIV-positive individuals-a clinical cohort study

OBJECTIVES

To measure the excess risk of cardiovascular disease (CVD) in HIV-positive individuals by comparing 'heart age' with real age and to estimate associations of patients' characteristics with heart age deviation (heart age-real age).

DESIGN

Clinical Cohort Study.

SETTING

Bristol HIV clinic, Brecon Unit at Southmead Hospital, Bristol, UK.

PARTICIPANTS

749 HIV-positive adults who attended for care between 2008 and 2011. Median age was 42 years (IQR 35-49), 67% were male and 82% were treated with antiretroviral therapy.

MAIN OUTCOME MEASURES

We calculated the Framingham 10-year risk of CVD and traced back to 'heart age', the age of an individual with the same score but ideal risk factor values. We estimated the relationship between heart age deviation and real age using fractional polynomial regression. We estimated crude and mutually adjusted associations of sex, age, CD4 count, viral load/treatment status and period of starting antiretroviral therapy with heart age deviation.

RESULTS

The average heart age for a male aged 45 years was 48 years for a non-smoker and 60 years for a smoker. Heart age deviation increased with real age and at younger ages was smaller for females than males, although this reversed after 48 years. Compared to patients with CD4 count <500 cells/mm(3), heart age deviation was 2.4 (95% CI 0.7 to 4.0) and 4.3 (2.3 to 6.3) years higher for those with CD4 500-749 cells/mm(3) and ≥750 cells/mm(3), respectively.

CONCLUSIONS

In HIV-positive individuals, the difference between heart age and real age increased with age and CD4 count and was very dependent on smoking status. Heart age could be a useful tool to communicate CVD risk to patients and the benefits of stopping smoking.

Cardiovascular disease and HIV infection

The emergence of chronic disease complications in controlled HIV disease has changed the landscape of HIV clinical care. HIV infection confers an increased cardiovascular disease risk, which is thought to be due to a complex interplay of mechanistic factors. While traditional cardiovascular risk factors likely play a role, recent evidence suggests that HIV-associated inflammation and immune activation are important mediators of cardiovascular risk. It is unclear whether established preventative interventions for the general population are applicable to HIV-infected patients, and the need to translate mechanistic knowledge into HIV-specific clinical interventions represents an important priority. Developing strategies to prevent cardiovascular disease in HIV-infected individuals calls for a multidisciplinary approach and represents an opportunity to exert a major public health impact in an at-risk population.

Cardiovascular risk assessment: a comparison of the Framingham, PROCAM, and DAD equations in HIV-infected persons

This study aims to estimate the risk of cardiovascular disease (CVD) and to assess the agreement between the Framingham, Framingham with aggravating factors, PROCAM, and DAD equations in HIV-infected patients. A cross-sectional study was conducted in an outpatient centre in Brazil. 294 patients older than 19 years were enrolled. Estimates of 10-year cardiovascular risk were calculated. The agreement between the CVD risk equations was assessed using Cohen's kappa coefficient. The participants' mean age was 36.8 years (SD = 10.3), 76.9% were men, and 66.3% were on antiretroviral therapy. 47.8% of the participants had abdominal obesity, 23.1% were current smokers, 20.0% had hypertension, and 2.0% had diabetes. At least one lipid abnormality was detected in 72.8%, and a low HDL-C level was the most common. The majority were classified as having low risk for CV events. The percentage of patients at high risk ranged from 0.4 to 5.7. The PROCAM score placed the lowest proportion of the patients into a high-risk group, and the Framingham equation with aggravating factors placed the highest proportion of patients into the high-risk group. Data concerning the comparability of different tools are informative for estimating the risk of CVD, but accuracy of the outcome predictions should also be considered.

Risk of coronary heart disease among HIV-infected patients: a multicenter study in Brazil

Cardiovascular disease has emerged as a crescent problem among HIV-infected population. This study aimed to determine the 10-year risk of coronary heart disease using the Framingham risk score among HIV-infected patients from three regions of Brazil. This is a pooled analysis of three cohort studies, which enrolled 3,829 individuals, 59% were men, 66% had white skin color, and mean age 39.0 ± 9.9 years. Comparisons among regions showed that there were marked differences in demographic, socioeconomic, clinical, and HIV-related characteristics. Prevalence of Framingham score ≥10 was 4.5% in the Southern, 4.2% in the Midwest, and 3.9% in the Northeast of Brazil. The Framingham score ≥10 was similar between regions for males, patients aged ≥60 years, with obesity, central obesity, hypertension, and diabetes mellitus. Women were three times more likely to have coronary heart disease in 10 years than men. Hypertension and diabetes increased more than four times the risk of coronary heart disease, followed by central obesity, obesity, and prehypertension. The use of antiretroviral agents and time since HIV diagnosis were not risk factors for coronary artery disease in 10 years. In conclusion, hypertension and diabetes are the strongest independent predictors of 10-year risk of coronary heart disease among HIV-infected population.

A literature review on cardiovascular risk in human immunodeficiency virus-infected patients: implications for clinical management

Introduction

In recent years, there has been growing concern about an increasing rate of cardiovascular diseases in human immunodeficiency virus-infected patients, which could be associated with side effects of highly active antiretroviral therapy. It is likely that the metabolic disorders related to anti-human immunodeficiency virus treatment will eventually translate into a increased cardiovascular risk in patients submitted to such regimens.

Objective

To evaluate if human immunodeficiency virus-infected patients receiving highly active antiretroviral therapy are at higher risk of cardiovascular diseases than human immunodeficiency virus infected patients not receiving highly active antiretroviral therapy, or the general population.

Research design and methods

We conducted a computer-based search in representative databases, and also performed manual tracking of citations in selected articles.

Result

The available evidence suggests an excess risk of cardiovascular events in human immunodeficiency virus-infected persons compared to non-human immunodeficiency virus infected individuals. The use of highly active antiretroviral therapy is associated with increased levels of total cholesterol, triglycerides, low-density lipoprotein and morphological signs of cardiovascular diseases. Some evidence suggested that human immunodeficiency virus-infected individuals on highly active antiretroviral therapy regimens are at increased risk of dyslipidemia, ischemic heart disease, and myocardial infarction, particularly if the highly active antiretroviral therapy regimen contains a protease inhibitor.

Conclusion

Physicians must weigh the cardiovascular risk against potential benefits when prescribing highly active antiretroviral therapy. Careful cardiac screening is warranted for patients who are being evaluated for, or who are receiving highly active antiretroviral therapy regimens, particularly for those with known underlying cardiovascular risk factors. A better understanding of the molecular mechanisms responsible for increased risk of cardiovascular diseases in human immunodeficiency virus-infected patients will lead to the discovery of new drugs that will reduce cardiovascular risk in human immunodeficiency virus-infected patients receiving highly active antiretroviral therapy.

Retinal arterioles narrow with increasing duration of anti-retroviral therapy in HIV infection: a novel estimator of vascular risk in HIV?

OBJECTIVES

HIV infection is associated with an increased risk of age-related morbidity mediated by immune dysfunction, atherosclerosis and inflammation. Changes in retinal vessel calibre may reflect cumulative structural damage arising from these mechanisms. The relationship of retinal vessel calibre with clinical and demographic characteristics was investigated in a population of HIV-infected individuals in South Africa.

METHODS

Case-control study of 491 adults ≥30 years, composed of 242 HIV-infected adults and 249 age- and gender-matched HIV-negative controls. Retinal vessel calibre was measured using computer-assisted techniques to determine mean arteriolar and venular diameters of each eye.

RESULTS

The median age was 40 years (IQR: 35-48 years). Among HIV-infected adults, 87.1% were receiving highly active antiretroviral therapy (HAART) (median duration, 58 months), their median CD4 count was 468 cells/µL, and 84.3% had undetectable plasma viral load. Unadjusted mean retinal arteriolar diameters were 163.67±17.69 µm in cases and 161.34±17.38 µm in controls (p = 0.15). Unadjusted mean venular diameters were 267.77±18.21 µm in cases and 270.81±18.98 µm in controls (p = 0.07). Age modified the effect of retinal arteriolar and venular diameters in relation to HIV status, with a tendency towards narrower retinal diameters in HIV cases but not in controls. Among cases, retinal arteriolar diameters narrowed with increasing duration of HAART, independently of age (167.83 µm <3 years of HAART vs. 158.89 µm >6 years, p-trend = 0.02), and with a HIV viral load >10,000 copies/mL while on HAART (p = 0.05). HIV-related venular changes were not detected.

CONCLUSIONS

Narrowing of retinal arteriolar diameters is associated with HAART duration and viral load, and may reflect heightened inflammatory and pro-atherogenic states of the systemic vasculature. Measurement of retinal vascular calibre could be an innovative non-invasive method of estimating vascular risk in HIV-infected individuals.