Norrie disease

Drug treatment might be contaminating factor

EDITOR--The burden on echocardiography services could indeed be reduced if natriuretic peptide concentrations plus electrocardiography were used as screening tools for left ventricular systolic dysfunction.[1] Nielsen et al's paper confirms the high negative predictive value of these tests. Concomitant drug treatment could, however, be a crucial contaminating factor.

The use of natriuretic peptides to diagnose left ventricular dysfunction in patients who are already taking cardiac drugs deserves particular attention. Diuretics, digoxin, and angiotensin converting enzyme inhibitors reduce natriuretic peptide concentrations.[2] Especially important is the fact that frusemide (furosemide) reduces these concentrations[3] but will have virtually no effect on an echocardiogram; it will not alter left ventricular dysfunction. Obviously, therefore, frusemide could severely distort the relation between natriuretic peptides and the echo finding of left ventricular systolic dysfunction.

The predictive value of natriuretic peptides could conceivably be considerably affected by the presence of frusemide and other cardiac drugs. This could explain why the sensitivity of natriuretic peptides is low in Nielsen et al's study. No study has yet addressed the usefulness of natriuretic peptides in identifying left ventricular dysfunction before diuretics have been prescribed, which is obviously the real clinical question. When general practitioners want to know if left ventricular dysfunction is the cause of breathlessness in a patient they want to be able to take a blood sample to measure the natriuretic peptide concentration there and then (and before prescribing a diuretic). They can then prescribe a diuretic as a failsafe mechanism pending the result of the test. The opposite may occur with [Beta] blockers as recent data suggest that they increase brain natriuretic peptide while having a beneficial effect on left ventricular dilatation.[4]

Cardiac drugs could therefore be a major contaminating factor in the use of natriuretic peptides to diagnose left ventricular dysfunction. An interesting question now arises from Nielsen et al's work: were there any differences in the predictive value of atrial natriuretic peptide concentration, clinical features, findings on electrocardiography, and heart rate and blood pressure between those patients taking cardiac drugs and those patients not taking any treatment in this study?

Robert Kelly research fellow

Allan D Struthers professor of clinical pharmacology Department of Clinical Pharmacology, Ninewells Hospital, Dundee DD1 9SY

[1] Nielsen OW, Hansen JF, Hilden J, Larsen CT, Svanegard J. Risk assessment of left ventricular systolic dysfunction in primary care. BMJ 2000;320:220-4. (22 January.)

[2] Murdoch DR, McDonagh TA, Byrne J, Blue L, Farmer R, Morton JJ, et al. Titration of vasodilator therapy in chronic heart failure according to plasma brain natriuretic peptide concentration. Am Heart J 1999;138:1126-32.

[3] Northridge DB, Newby DE, Rooney E, Norrie J, Dargie HJ. Comparison of the short-term effects of candoxatril, an orally active neutral endopeptidase inhibitor, and frusemide in the treatment of patients with chronic heart failure. Am Heart J 1999;138:1149-57.

[4] RESOLVD Investigators. Effects of metoprolol CR in patients with ischemic and dilated cardiomyopathy. Circulation 2000; 101:378-84.

Authors' reply

EDITOR--Kelly and Struthers make a point that applies in many branches of medical diagnosis--namely, that drug treatment may influence not only subjective symptoms but also biological disease markers. If a drug breaks into a complicated system of feedback regulations, as is the case with natriuretic peptides, the net effect is unpredictable and must be assessed empirically. The question whether cardiac medication influences the predictive power of natriuretic peptides in subjects with minor symptoms is therefore an important one.

In our study we applied the tests on a broad spectrum of suspected heart patients from general practice. We did not consider treatment in the analysis because diuretic treatment is based on subjective and arbitrary decisions, because a subdivision of the small cohort would increase the risk of making type II errors, and because we only had limited space for publication.

Despite these limitations, and prompted by the query by Kelly and Struthers, we have made a renewed analysis of those 25 patients treated with either loop diuretics or an angiotensin converting enzyme inhibitor in combination with another diuretic, comparing them with 95 patients without this treatment (table). Missing values and pacemaker patients were excluded. Totals are thus different from those of the original table 3. The prevalences of left ventricular systolic dysfunction were 30% and 7% respectively.

This difference in prevalence of systolic dysfunction makes it hard to compare test behaviour in the two groups. It appears from the table, however, that electrocardiographic anomaly gives full sensitivity (scoring no false negatives), whereas the predictor heart rate [is greater than] diastolic blood pressure has an unchanged performance. As Kelly and Struthers may have expected, the natriuretic peptide improves its discriminative power in the untreated patient group, especially if the cut-off point is lowered. The lower cut-off point, however, weakens the predictive positive value of the test.

In conclusion, there seem to be fewer false negatives among the untreated patients by several criteria. The table therefore suggests that when testing is restricted to untreated patients one can more safely rule out left ventricular systolic dysfunction by normal electrocardiographic results and a normal natriuretic peptide concentration. The table also suggests that echocardiography should always be considered if a loop diuretic is required to control symptoms. Since this subanalysis is based on a small patient sample, we plan to undertake a more careful analysis in a larger cohort, where potential confounders other than drug treatment will also be considered.

Risk assessment of left ventricular dysfunction in patients according to treatment with diuretics (loop diuretic or angiotensin converting enzyme inhibitor plus other diuretic)

Olav Wendelboe Nielsen research registrar own@dadlnet.dk Cardiovascular Department Y, Copenhagen University Hospital, 2400 Bispebjerg, Denmark

Jorgen Hilden lecturer Department of Biostatistics, Copenhagen University, 2200 Panum Institute, Denmark

Jens Svanegaard chief physician Department of Cardiovascular Medicine, Haderslev Hospital, Denmark

Jorgen Fischer Hansen chief consultant Cardiovascular Department Y, Copenhagen University Hospital, 2400 Bispebjerg

COPYRIGHT 2000 British Medical Association
COPYRIGHT 2000 Gale Group