Article Text

Download PDFPDF

Screening for left ventricular dysfunction: a step too far?
  1. T A McDonagh
  1. Correspondence to:
    Dr Theresa A McDonagh, Department of Medical Cardiology, University of Glasgow, Glasgow Royal Infirmary, 10 Alexandra Parade, Glasgow G31 2ER, UK;

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Much has been written and said about the burden of heart failure in the community, and the need for its accurate and rapid diagnosis to facilitate access to the disease modifying therapies which can alleviate both the morbidity and mortality which accompany the syndrome. At this time, however, it might seem that we are overwhelmed with the problem of dealing with the manifest phase of this condition, never mind trying to address the question posed in the title of this article. Nevertheless, it is my remit to step back further in the natural history of heart failure to try and answer the question of whether we should be screening for an asymptomatic precursor of heart failure.

Before embarking on this, it is worth reminding ourselves of the criteria which should be fulfilled before we screen for a disease. Namely, the disease we are seeking to detect should be common, it should be serious (that is, represent a significant public health problem), it should have a recognisable latent phase, it must be treatable, and lastly there should be a cost effective and acceptable test with which to detect it.1 So lets apply these criteria to heart failure.

Firstly, is there a recognisable latent phase? The most common antecedent of chronic heart failure (CHF) and the one we understand best is left ventricular systolic dysfunction (LVSD) although a proportion of heart failure cases arise in the presence of normal systolic function. As we have only recently begun to unravel how to treat heart failure caused by systolic dysfunction and know very little about how to treat the latter condition effectively, it would be wise to contain this argument for now to the scenario of heart failure arising from systolic dysfunction. Pertaining to this, we are now becoming increasingly aware that many patients who go on to develop the syndrome of heart failure pass through a phase of varying length where they have significant left ventricular damage but lack the classical symptoms and signs of heart failure: the precursor phase of heart failure—asymptomatic left ventricular dysfunction (ALVD).


Epidemiological work in North Glasgow showed that significant LVSD was present in 2.9% of the population aged 25–74; just under half of this was asymptomatic giving a population prevalence in this age group of 1.4%.2 That prevalence rose with age from zero in those under 35 years old to 3.2% in both males and females over 65 years. Other studies have also confirmed a relatively high prevalence of this condition: 5.2% of men and 1.9% of women, free of previous myocardial infarction, in a recent study from Framingham, had an asymptomatic decline in their fractional shortening.3 In a more elderly US population, Gardin and colleagues reported that 6.8% of asymptomatic men and 4.2% of women had a left ventricular ejection fraction ≤ 35%.4 Similarly in Europe, in the Rotterdam study, in 2832 participants over 55 years, 2.2% had an an asymptomatic reduction in their fractional shortening.5 And most recently in Poole, UK, Morgan and colleagues reported that 7.7% of the population had left ventricular dysfunction—52% was previously undiagnosed.6 So a fairly consistent pattern is emerging that between 3–5% of the population, depending on age, have ALVD. This might seem a bit daunting. However, like CHF itself, ALVD resides in recognisable high risk groups; in North Glasgow 71% had some evidence of previous ischaemic heart disease (IHD), and 38% had the combination of IHD and hypertension.2


We are very well aware of the poor prognosis associated with the syndrome of CHF,7 in addition to its effect in reducing quality of life to a greater extent than in most other chronic medical conditions.8 It is also expensive, with 20–30% of patients with CHF being hospitalised at least once per annum; this leads to a total expenditure amounting to 1–2% of the total National Health Service (NHS) budget, most of which is attributable to hospitalisation.9 We also have good data from the SOLVD (studies of left ventricular dysfunction) prevention study showing that the four year mortality of ALVD on enalapril was 14.8%,10 and from the SAVE (survival and ventricular enlargement) trial indicating that the four year mortality in post-myocardial infarction ALVD was 20% on the active treatment (captopril).11 We now also have evidence from population based studies on the outcome of ALVD. In the Framingham offspring cohort, where they studied 1483 men free of cardiovascular disease, the combination of a reduced fractional shortening and an increased left ventricular end diastolic dimension on echocardiography conferred a hazard ratio of 3.77 for the subsequent development of cardiovascular disease.12 More importantly, in Framingham’s cohort of 4744 men and women free of myocardial infarction at baseline, every one standard deviation increment in the left ventricular end diastolic dimension was associated with a hazard ratio of 1.47 for the development of CHF over the next 11 years.3 Most recently, we also have evidence from the North Glasgow MONICA (monitoring trends and determinants in cardiovascular disease) study indicating that the four year mortality of significant left ventricular dysfunction was 21%; there was no difference between symptomatic and asymptomatic subjects, although the numbers in each group were small.13 It seems, therefore, that ALVD in the community, as in the clinical trials, is not a benign condition.


There is certainly good evidence now from the SAVE, SOLVD prevention, and TRACE (trandolapril cardiac evaluation) studies that ACE inhibitors reduce the development of heart failure.10,11,14 In the SOLVD prevention study enalapril reduced hospitalisations for CHF by one third, and in the SAVE study there was a 22% reduction with captopril.10,11 Regarding mortality, in the SAVE study there was a significant reduction in all cause mortality associated with angiotensin converting enzyme (ACE) inhibitor use, and in SOLVD there was a reduction in the combined end point of both hospitalisation for heart failure and mortality. Most recently evidence has also emerged for the use of β adrenoreceptor antagonists in heart failure, with the CAPRICORN study demonstrating that carvedilol reduces all cause mortality in post-myocardial infarction left ventricular dysfunction, be it symptomatic or asymptomatic.15 It seems, therefore, that we do have effective treatment.


That leaves us with the question of the availability of a cost effective and acceptable test for ALVD. Obviously echocardiography provides us with an acceptable non-invasive test for LVD. However, few would advocate widespread population screening by echocardiography. Detailed echocardiography is expensive and, therefore, unlikely to be a cost effective solution. It seems also that our echocardiography services are overburdened trying to cope with the symptomatic phase of this epidemic, never mind trying to detect latent disease. We are more realistically looking at ways of filtering out subjects who require to proceed to echocardiography. A potential means of achieving this is via a routine 12 lead ECG, as a normal ECG has been shown in subjects with CHF to indicate that the presence of significant left ventrciular dysfunction is unlikely.16 However, it has not been tested in the asymptomatic population and there seems to be little enthusiasm in primary care for using the 12 lead ECG in this manner. A biochemical test is undoubtedly the preferred option.

Of all the putative biochemical markers the natriuretic peptide hormones have shown the most potential and are the most studied. Both the active forms of atrial natriuretic peptide (ANP) and brain natriuretic peptide (BNP) and their inactive N terminal fragments (N-ANP and NT-pro-BNP) have been known for some time to be elevated in their circulating concentrations in patients with CHF and ALVD.17–19 Recent work from a population based study in 1252 subjects aged 25–74 years also demonstrated increases in the plasma concentrations of both BNP and N-ANP in subjects from the community with ALVD.20,21 On receiver operator characteristic (ROC) analysis, BNP appeared to have superior ability to N-ANP for the detection of left ventricular dysfunction in the population. Indeed both BNP and the N-terminal of BNP were the strongest independent predictors of left ventricular dysfunction in the population—the other significant ones being advanced age, male sex, and the presence of IHD. When we turn to the detection of ALVD in the community, BNP also performs better than ANP or N-ANP.

Clearly it would not be cost effective to screen the entire population for ALVD. We know where the disease lies—in high risk subjects who are older, with evidence of IHD, hypertension or diabetes. In the North Glasgow MONICA study, when screening with BNP was targeted at those high risk subjects over the age of 55 years with some evidence of prior IHD, the overall accuracy of the test, as measured by the area under the ROC curve (AUC), was 0.88.20 This compares favourably with prostate specific antigen in screening for prostatic carcinoma (AUC 0.94),22 and is superior to mammography for breast carcinoma (AUC 0.85)23 and Papanicolau smears for cervical carcinoma (AUC 0.70).24 Another population based study has also confirmed that BNP has acceptable accuracy to detect LVD in the elderly population.6 These studies extend earlier work in the post-myocardial infarction situation, where BNP appeared to be the most sensitive marker in detecting a left ventricular ejection fraction ≤ 40% post-myocardial infarction in comparison to various clinical scoring systems, as well as ANP.25 As in the assessment of CHF, so too in the detection of ALVD the strength of BNP is in the exclusion of left ventricular dysfunction. It has an extremely high negative predictive accuracy (98–99%) meaning that low values of BNP virtually exclude significant systolic dysfunction. High values do not equate solely to the presence of left ventricular dysfunction.

To date, therefore, we have shown that ALVD broadly meets the criteria which need to be fulfilled before we screen for a condition. Natriuretic peptides, in particular BNP and NT-pro-BNP, have the accuracy to detect ALVD and their accuracy can be improved by targeting groups at high risk of ALVD. However, is screening for ALVD just a step too far at the moment?


Undoubtedly some fundamental questions still need to be answered. We have still to demonstrate that it will be cost effective to screen for ALVD. This is a complex question involving not only the cost of the test but also the efficacy and cost of the resultant treatment for the detected left ventricular dysfunction. Cost effectiveness is unproven but likely. We know that the detection and treatment of left ventricular dysfunction in the SAVE study was cost effective.26 In addition detailed echocardiography is likely to be more expensive than natriuretic peptide assays. However, we still await a definitive study on cost effectiveness.

The ease of use of potential peptide assays is also likely to be crucial if any screening programme was to be undertaken. Many of the clinical and population based studies quoted here used time consuming radioimmunoassays requiring prior plasma extraction. However, rapid assays for both BNP and NT-pro-BNP have recently become available and will facilitate the progress of peptide assays into clinical practice as well as reducing cost.

We also need to know more about the performance of BNP/NT-pro-BNP assays in the screening arena before they can be adopted widely. What are the cut points we should use to determine abnormality? This is especially poignant as preliminary work from several peptide research groups have demonstrated that the plasma concentrations rise with age and are different in men and women.27 In addition, it should be reiterated that an increased plasma BNP/NT-pro-BNP (though highest in subjects with left ventricular dysfunction) is not just indicative of systolic dysfunction but rather of cardiac dysfunction, including the entire spectrum of structural and functional abnormalities, as well as of renal dysfunction which leads to reduced clearance of the natriuretic peptides.28 As such, finding a high BNP/NT-pro-BNP would necessitate further cardiorenal investigation, including echocardiography.


Although screening for left ventricular dysfunction is now technically possible using natriuretic peptides, before it can be implemented we need data on cost effectiveness and detailed work to devise appropriate care pathways for those found to have abnormal values. However, it is still very much on the agenda as a tangible way of detecting and treating heart failure at a much earlier point in its natural history than at the more advanced stages when patients currently present to medical care.


Question: Do you have any idea of the aetiology of the asymptomatic left ventricular dysfunction? Is it nearly all ischaemic heart disease or is there something interesting going on with the cardiomyopathies and other things?

Dr McDonagh: In the UK studies it seems to be predominantly ischaemic heart disease in about 75% of cases, the rest of it hypertension. There are obviously a few odd cases of dilated cardiomyopathy but not a huge number.

Question: You didn’t mention left ventricular diastolic dysfunction. Are there any studies looking at biochemical markers in that group of people?

Dr McDonagh: There are one or two epidemiological studies that are as yet unpublished but they have been presented. Patients with heart failure and preserved systolic function do have higher levels of natriuretic peptides.

Question: Did you look at your population to see what the prevalence of diastolic dysfunction was?

Dr McDonagh: We looked at the prevalence of diastolic filling abnormalities by echo in our population—E:A ratios, deceleration times, and various combinations thereof—and we found that about 2–3% of the population did have these diastolic filling abnormalities. What we could not demonstrate was that this was in any way indicative that these people had a higher mortality or had any more symptoms. I’m not sure that we were measuring anything biologically important in them or, indeed, whether we were measuring true relaxation of the ventricle.

Question: Do you know anything about the mode of death of patients with asymptomatic left ventricular dysfunction? Are there a lot of patients who die suddenly and unexpectedly before they ever develop symptomatic heart failure?

Dr McDonagh: There is good evidence from the clinical trials that the predominant mode of death in asymptomatic left ventricular dysfunction is sudden death as opposed to progressive heart failure. In the Glasgow study, I didn’t want to do it by cause of death and mode of death as we had so few deaths, it wouldn’t have been very scientific, but more of them were sudden deaths before patients developed symptoms, in keeping with the clinical trials.