Correspondence

Multiple Biomarkers and Cardiovascular Risk

N Engl J Med 2008; 359:760-761August 14, 2008

Article

To the Editor:

Zethelius et al. (May 15 issue)1 suggest that the combination of multiple biomarkers with established risk factors results in an improved assessment of cardiovascular risk among elderly men. Unfortunately, they did not validate their findings in an independent cohort of patients.2 Thus, a possibly spurious association they found in a highly selected subgroup may be erroneously generalized, particularly to women and to other age and ethnic groups. Why should we spend about €35 ($55) per patient to optimize our prognostic evaluation without having any proven effect on individual risk reduction? In contrast, simple and cheap risk scores (e.g., the Framingham score and the Systematic Coronary Risk Evaluation [SCORE]) are available and involve treatable risk factors. However, these risk scores are rarely calculated on a routine basis, so risk factors in such patients are often untreated despite the proven benefits of intervention.3 Finally, we understand that the completeness of the Swedish Cause of Death Register is estimated to exceed 99%.4 However, did the authors account for the time span between deaths and the reporting of those deaths to the register?

Roman Pfister, M.D.
Martin Hellmich, D.Med.Sc.
University of Cologne, 50924 Cologne, Germany
martin.hellmich@uni-koeln.de

4 References

1. 1

   Zethelius B, Berglund L, Sundstrom J, et al. Use of multiple biomarkers to improve the prediction of death from cardiovascular causes. N Engl J Med 2008;358:2107-2116
   Full Text | Web of Science | Medline

2. 2

   Ransohoff DF. Rules of evidence for cancer molecular-marker discovery and validation. Nat Rev Cancer 2004;4:309-314
   CrossRef | Web of Science | Medline

3. 3

   Hall LM, Jung RT, Leese GP. Controlled trial of effect of documented cardiovascular risk scores on prescribing. BMJ 2003;326:251-252
   CrossRef | Web of Science | Medline

4. 4

   Thalib L, Wedren S, Granath F, et al. Breast cancer prognosis in relation to family history of breast and ovarian cancer. Br J Cancer 2004;90:1378-1381
   CrossRef | Web of Science | Medline

To the Editor:

Zethelius et al. describe the use of biomarkers for disease screening in a cohort of elderly men. Coexisting illnesses in the elderly, such as chronic liver disease, have a substantial influence on levels of such biomarkers. Many patients with chronic liver disease have coexisting cardiac disease.1 In patients with chronic liver disease whom we have identified, cardiac disease has a significant effect on survival, regardless of the cause.2

Our preliminary studies have shown that levels of brain natriuretic peptide (BNP) correlate well with markers of hepatic synthetic function and the severity of liver disease.3 More severe disease (a score of >17 on the Model for End-Stage Liver Disease [MELD] scale or of >11 on the Child–Pugh scale) is more closely related to BNP levels than is less severe disease. Elevations in BNP levels alongside hepatic synthetic dysfunction do not appear to relate solely to structural cardiac disease. Moreover, C-reactive protein (CRP) is synthesized in the liver and is influenced by hepatic reserve and is quantitatively unreliable in patients with cirrhosis. For these reasons, we would suggest caution when using biomarkers to screen elderly patients for chronic disease. Abnormalities may be explained by undiagnosed coexisting disease, such as chronic liver disease.

Mark J. Austin, M.R.C.P.
Michael A. Heneghan, M.D.
King's College Hospital, London SE5 9RS, United Kingdom

3 References

1. 1

   Pozzi M, Carugo S, Boari G, et al. Evidence of functional and structural cardiac abnormalities in cirrhotic patients with and without ascites. Hepatology 1997;26:1131-1137
   Web of Science | Medline

2. 2

   Austin MJ, Bernal W, Wendon JA. A simple scoring system accurately predicts outcomes in patients with chronic liver disease admitted to a liver intensive care unit. Hepatology 2006;44:Suppl:446A-446A
   Web of Science

3. 3

   Austin MJ, Sizer L, Langworthy R, Sherwood R, Wendon J, Heneghan MA. The role of B type natriuretic peptide (BNP) as a marker of structural cardiac disease in patients with cirrhosis. Hepatology 2007;46:594A-595A
   Web of Science

Author/Editor Response

Pfister and Hellmich express concern regarding the lack of validation, the unknown generalizability, and the potential cost–benefit of measuring a combination of multiple biomarkers in clinical practice for the prediction of cardiovascular disease. Although we acknowledge such limitations in our article, we are grateful for the opportunity to once more clarify our viewpoint on this matter. Before these diagnostic biomarkers for risk prediction are used in clinical practice, it is essential that our findings be validated in other study populations. The cost–benefit of such an approach and the potential benefits of an intervention on the basis of the test results need to be thoroughly evaluated.

The completeness of ascertainment and accuracy of classification in the Swedish Cause of Death Register represent a particular strength of our study.1 The time span between death and the availability of data in this nationwide compulsory register is approximately 4 months. The end of our follow-up was December 31, 2003. We retrieved the outcome data from the register in May 2005. Thus, it is unlikely that the time span between death and reporting of death to the register influenced our results. Any potential misclassification would most likely drive the associations toward the null hypothesis.

Austin and Heneghan propose the excellent hypothesis that the use of CRP and N-terminal prohormone BNP (NT-proBNP) to predict the risk of death from cardiovascular causes could have been confounded by coexisting liver damage. We addressed this issue by excluding 169 patients with potential hepatic parenchymal damage (which we defined as a serum aspartate aminotransferase level of >0.60 μkat per liter or a serum alanine aminotransferase level of >0.80 μkat per liter) or signs of hepatic synthetic dysfunction (a serum albumin level of <37 g per liter). Our results remained essentially unaltered in this subgroup, even after the addition of the liver-disease markers to the multivariable model with established cardiovascular risk factors (multivariable hazard ratio for 1-SD increase in CRP, 1.42; 95% confidence interval [CI], 1.16 to 1.73; P<0.001; multivariable hazard ratio for 1-SD increase in NT-proBNP, 2.06; 95% CI, 1.71 to 2.48; P<0.001). Thus, our data argue against confounding by overt or subclinical liver damage as an explanation for our findings. The predictive value of these biomarkers in patients with chronic liver disease needs to be investigated in future studies.

Björn Zethelius, M.D., Ph.D.
Per Venge, M.D., Ph.D.
Johan Ärnlöv, M.D., Ph.D.
Uppsala University, 75185 Uppsala, Sweden
johan.arnlov@pubcare.uu.se

1 References

1. 1

   Tunstall-Pedoe H, Kuulasmaa K, Amouyel P, Arveiler D, Rajakangas AM, Pajak A. Myocardial infarction and coronary deaths in the World Health Organization MONICA Project: registration procedures, event rates, and case-fatality rates in 38 populations from 21 countries in four continents. Circulation 1994;90:583-612
   Web of Science | Medline

Citing Articles (1)

Citing Articles

1. 1

   Kyung-Hoon Lee, Jang-Young Kim, Sang-Baek Koh, Seung-Hwan Lee, Junghan Yoon, Sang-Woo Han, Jong-Ku Park, Kyung-Hoon Choe, Byung-Su Yoo. (2010) N-Terminal Pro-B-type Natriuretic Peptide Levels in the Korean General Population. Korean Circulation Journal 40:12, 645
   CrossRef