Rapid Responses to:

RESEARCH: Wouter de Ruijter, Rudi G J Westendorp, Willem J J Assendelft, Wendy P J den Elzen, Anton J M de Craen, Saskia le Cessie, and Jacobijn Gussekloo Use of Framingham risk score and new biomarkers to predict cardiovascular mortality in older people: population based observational cohort study BMJ 2009; 338: a3083 [Abstract] [Full text]

Rapid Responses: Submit a response to this article

Rapid Responses published:

Fatty Acid Profiles, major factor, were omitted

Edward N. Siguel   (20 January 2009)

Accurate and Ethical?

Chris C Gunstone, Burton, Staffs DE14 2JA   (24 January 2009)

Does the ability of homocysteine to predict mortality of the very elderly merely reflects their reduced renal function?

Michael Bursztyn, Iddo Z Ben-Dov   (26 January 2009)

Authors' Reply

Wouter de Ruijter, Rudi G.J. Westendorp, Willem J.J. Assendelft, Wendy P.J. den Elzen, Anton J.M. de Craen, Saskia le Cessie and Jacobijn Gussekloo.   (29 January 2009)

New markers predicting mortality in the elderly

Yoram Maaravi, Jeremy M Jacobs, Jochanan Stessman   (11 February 2009)

What is the cut-off value of homocysteine that can be considered as biomarker?

Neeru Gupta, A.K. Mathur. Suwarna, Nivedita Gupta.   (14 September 2009)

Fatty Acid Profiles, major factor, were omitted 20 January 2009
Edward N. Siguel, PHysician, lipid research essentiial fats, MD, USA, 20898 Send response to journal: Re: Fatty Acid Profiles, major factor, were omitted	Predictor models are functions; Dependent variable = = f(independent variables). Some functions allow relationships to be explicit and clearly understood, such as multivariate regression (not necessarily linear). In a multivariate regression, the predictive value of the equation is (in part) measured by the R square. Writers often misunderstand R square. It is possible to have high predictive value with very low R square (example = circle). However, low R squares usually means the model is a poor predictor. When the R square is below .3 (as it is in most models used to predict cardiovascular disease risk, and as I believe it is with Framingham models), it means that the most important predictor factors have been overlooked. The methods of communication theory, such as area under receiver operating characteristic curve, must be used with great caution when extrapolated to mean predictive value for cardiovascular disease risk. Communication theory deals with stochastic processes with emphasis on error detection and correction, and predicting the value of a signal transmitted. Receiver operating characteristic curve may be used to analyze the value of comparable tests. However, for predicting risk, authors should at least include a model that more directly reflects the intended purposes, such as multivariate regression or equivalent mathematical functions. Otherwise, the data and interpretation can be misleading. Incidentally, p values are not relevant; with a large sample size, even small differences become statistically significant. There is substantial agreement today that fatty acid profiles and levels of w6s and w3s are significant factors in cardiovascular disease. I invented a method to accurately measure fatty acids in humans and showed that subjects participating in the Framingham Heart Study had substantial deficiencies and imbalances of essential fats (w3s and w6s). I also showed that fatty acid profiles are excellent predictors of cardiovascular disease and risk, likely better predictors than any other known factor. Levels of essential fats are strongly determined by diet and mildly affected by physiological conditions. In contrast, CRP levels are highly variable and strongly affected by mild physiological conditions. Levels of essential fats can be easily modified by diet; levels of CRP are tricky to modify by external factors and there is a dispute on whether they are a cause or a consequence of cardiovascular disease. Levels of essential fats are not a consequence of cardiovascular disease. The authors did not report fatty acid profiles. I propose it makes far more sense to accurately measure fatty acid profiles and/or adjust diets to optimize fatty acid profiles as I have proposed in my papers, and to ignore CRP except, perhaps, for research purposes. From a clinical perspective, I would rather find out the dietary intake of essential fats (relatively easy to do if one asks appropriate questions), and prescribe an optimal diet, than find out CRP levels. Siguel E, Maclure, M. Relative Activity of Unsaturated Fatty Acid Metabolic Pathways in Humans. Metabolism, 1987; 36: 664-69. Siguel, E. Method and Apparatus for Diagnosis of Fatty Acid or Lipid Abnormalities. U.S. Patent No. 5075101, Issue date 12/24/91. Siguel E, Lerman, RH. Altered Fatty Acid Metabolism in Patients With Angiographically Documented Coronary Artery Disease. Metabolism 1994; 43:982-93. Siguel E. Essential and Trans Fatty Acid Metabolism in Health and Disease. Nutrition Issue. Comprehensive Therapy, 1994; 20(9):500-10 (Review). Siguel, E. A New Relationship between Total/HDL Cholesterol and Polyunsaturated Fatty Acids. Lipids, 1996; 31:S51-6. Edward Siguel, MD, PhD. www. Essentialfats.com (see statement on potential financial conflicts). Author has patent on fatty acid profiles, reported above. Competing interests: Own patent for fatty acid analysis. Research on fatty acids.
Accurate and Ethical? 24 January 2009
Chris C Gunstone, GP 72 Gordon Street, Burton, Staffs DE14 2JA Send response to journal: Re: Accurate and Ethical?	The authors proclaim that their homocysteine prediction model accurately predicts cardiovascular mortality - yet whilst it is an improvement on Frammingham which predicted 12 out of 35 deaths, 20 out of 35 predicted by their model does not seem very accurate! The ethical aspect is that I assume that if a person is identified at high risk, one would then be offering lifestyle advice, plus pharmacological interventions - statins, aspirin etc. The patient needs to make an informed consent to treatment. Whilst we may be able to say that treatment might reduce their risk of a cardiovascular death (often quite a quick way to die), if they do not die of a heart attack, what will they die of? The longer one lives the greater the risk of dying with cancer and dementia. In Britain the life expectancy of an 85 year old male is just over 9 years, a female would expect almost 11. Should we be discussing with our patients how they would like to die? Do we know that cardiovascular interventions will extend life or improve quality of life? All our patients are going to die. For many the mode of dying is important. Just because we have the means to prevent a disease, it does not necessarily mean that it is in our patients best interests that we embark on treatment. The patient needs the information to enable them to make informed decisions. Does thie information exist? Competing interests: None declared
Does the ability of homocysteine to predict mortality of the very elderly merely reflects their reduced renal function? 26 January 2009
Michael Bursztyn, Professor of Medicine Hadassah-Hebrew University Medical Center, Mount-Scopus, 91240 Jerusalem, Israel, Iddo Z Ben-Dov Send response to journal: Re: Does the ability of homocysteine to predict mortality of the very elderly merely reflects their reduced renal function?	de Ruijter et al, suggest that the Framingham score does not perform well as a cardiovascular mortality predictor in elderly subjects in their eighties. They also found that plasma homocysteine, has a cardiovascular mortality predictive power. Nevertheless the authors did not take into account renal function, which is a major determinant of mortality in the elderly in many studies including one of our own (1). As homocysteine levels relate reciprocally to renal function (2) a finding confirmed by a metaanalysis of 27,000 patients (3), the predictive power de Ruijter et al, have found may reflect that of reduced glomerular filtration rate in older subjects. Indeed in after taking renal function into account homocysteine may loose its predictive power (4,5). Moreover, the lack of predictive power of folic acid (a determinant of homocysteine levels) also speaks somewhat against a role of homocysteine as a mediator. Indeed it may well be that homocysteine is no more than an "expensive creatinine"(6) in this context. References 1) Maaravi Y, Bursztyn M, Rozenberg-Hammerman R, Stessman J. Glomerular filtration rate estimation and mortality in an elderly population. Q J Med 20074;100:441-449. 2) Friedman AN, Bostom AG, Selhub J, Levey AS, Rosenberg IH: The kidney and homocysteine metabolism. J Am Soc Nephrol 2001;12:2181 –2189. 3) Kielstein JT, Salpeter SR, Buckley NS, Cooke JP, Fliser D. Two cardiovascular risk factors in one? Homocysteine and its relation to glomerular filtration rate. A meta-analysis of 41 studies with 27,000 participants.Kidney Blood Press Res. 2008;31:259-67. 4) Friedman AN, Hunsicker LG, Selhub J, Bostom AG; Collaborative Study Group. Total plasma homocysteine and arteriosclerotic outcomes in type 2 diabetes with nephropathy.J Am Soc Nephrol. 2005;16:3397-402. 5) Menon V, Sarnak MJ, Greene T, Wang X, Pereira AA, Beck GJ, Kusek JW, Selhub J,Collins AJ, Levey AS, Shlipak MG. Relationship between homocysteine and mortality in chronic kidney disease.Circulation. 2006;113:1572-7. 6) Bostom AG: Homocysteine: "Expensive creatinine" or important modifiable risk factor for arteriosclerotic outcomes in renal transplant recipients? J Am Soc Nephrol 2000;11 : 149 –151, Competing interests: None declared
Authors' Reply 29 January 2009
Wouter de Ruijter, general practitioner-clinical researcher Leiden University Medical Center, Dept. of Public Health and Primary Care, 2300RC Leiden (NL), Rudi G.J. Westendorp, Willem J.J. Assendelft, Wendy P.J. den Elzen, Anton J.M. de Craen, Saskia le Cessie and Jacobijn Gussekloo. Send response to journal: Re: Authors' Reply	In his Rapid Response of 26 January, Bursztyn (Jerusalem, Israel) questions the predictive value of homocysteine that we observed in our study in the oldest old, indicating that we did not take renal function into account. Furthermore, he states that renal function is a major determinant of mortality in the elderly, and that homocysteine levels are reciprocally associated with renal function. Therefore, Bursztyn hypothesizes, homocysteine may loose its predictive power after taking renal function into account, thus becoming no more than ‘expensive creatinin’. Indeed, in our study, median homocysteine levels varied from 13.1 micromol/L in the lowest tertile of creatinin clearance to 11.2 micromol/L in the highest tertile (Jonckheere-Terpstra p<0.001). However, there was no significant association between creatinin clearance and 5-year all cause mortality or cardiovascular mortality. This also applies to the full cohort of the Leiden 85-plus Study (n=599). So when we entered creatinin clearance into the homocysteine-based model, the 5-year cardiovascular mortality risk in the high-risk versus the low-risk category remained unchanged (RR crude model 3.4, 95%CI 1.4 to 8.1 and RR adjusted model 3.3, 95%CI 1.4 to 7.9). In other words, additional correction for renal function did not influence our results with regard to the predictive value of homocysteine. We addressed this in a small sentence in the results section (page 4: ‘This risk did not change after adjustment for creatinin clearance’), but did not explicitly discuss the matter thereafter. Competing interests: None declared
New markers predicting mortality in the elderly 11 February 2009
Yoram Maaravi, Senior Physician, Department of Geriatrics and Rehabilitation Hadassah Hebrew-University Medical Center, Mount Scopus, Jerusalem 91240, ISRAEL, Jeremy M Jacobs, Jochanan Stessman Send response to journal: Re: New markers predicting mortality in the elderly	Dear Sir: We read with great interest the recent paper by de Ruijter et al. (1), which reported that classic risk factors do not predict cardiovascular mortality among a sample of community dwelling old people, , whilst new biomarkers, namely homocysteine concentration, do. The authors should be congratulated for addressing the important yet unresolved issue of mortality prediction in the old population. We wish to report preliminary results in support of an additional novel predictor of mortality among community dwelling older people: low- normal thyrotropin levels. The significance of subclinical thyroid dysfunction in the aged is still controversial (2), yet most studies derive "normal" thyroid function tests from young populations and the definition of normality in old age, whether statistical or epidemiologic, is still lacking (3). We have recently analyzed the relation of blood thyrotropin level to mortality in a community-dweling elderly population as part of the Jerusalem Longitudinal Study (4). The study population consisted of 473 subjects all 77 years of age, with a thyrotropin value in the "normal" range (0.35-5.5 mIU/L). None of the subjects suffered from thyroid disease or received medications interfering with thyroid function tests. Subjects were stratified into quartiles according to their thyrotropin level: 0.35-1.11 mIU/L, 1.11-1.65 mIU/L, 1.65-2.43 mIU/L, 2.43-5.5 mIU/L. Mortality after 12 years was highest for subjects in the lowest quartile (0.35-1.1 mIU/L), 55%, p=0.039. This mortality risk was found to be significant after controlling for gender, hypertension, diabetes mellitus, malignant diseases, economic status and self rated health in a Cox proportional hazards model (HR 1.54, 95% CI 1.0-2.4, p=0.049). We propose that normal reference values for thyrotropin in old people be validated and suggest that low-normal thyrotropin levels be considered as an independent mortality risk. 1) De Ruijter W, Westendorp RGJ, Assendelft WJJ et al. Use of Framingham risk score and new biomarkers to predict cardiovascular mortality in older people: population based observational cohort study. BMJ 2009; 338a:3083. 2) Ladenson PW. Cardiovascular consequences of subclinical thyroid dysfunction: more smoke but no fire. Ann Intern Med 2008; 148:880-881. 3) Scobbo RR. Thyroid status and survival in old age. JAMA 2005; 293:1447. 4) Stessman J, Cohen A, Ginsberg GM et al. The Jerusalem seventy year olds longitudinal study: I. Description of the initial cross-sectional survey. Eur J Epidemiol 1995; 11:675-84. Competing interests: None declared
What is the cut-off value of homocysteine that can be considered as biomarker? 14 September 2009
Neeru Gupta, Deputy Director General and Scientist E Indian Council of Medical Research, Ansari Nagar, New Delhi-110029., A.K. Mathur. Suwarna, Nivedita Gupta. Send response to journal: Re: What is the cut-off value of homocysteine that can be considered as biomarker?	Leiden 85-plus study is a very interesting study, but some queries remain unanswered. Receiver Operating characteristic curve though showed 100% sensitivity and 100% false positivity (1-specificity=1), which is a good characteristic to be considered a biomarker. But, still authors have not given the cut-off values for the biomarkers, particularly homocysteine, beyond which it will be considered a risk factor. Homocysteinemia can occur due to deficiency of Vitamin B6, B12 and Folic acid so their deficiencies should have been taken as exclusion criteria in the study, to consider homocysteinemia as the sole bio-marker of the study. Body Mass Index should have been matched or defined as it would be a proxy indicator for nutrition status (as different nutritional status will have different levels of homocysteinemia) and anemia (especially megaloblastic or macrocytic anemia) should have been excluded to exclude B12 and Folic acid deficiency. Either the aforesaid exclusion criteria should have been adhered to or the confounding of these factors should have been adjusted before considering homocysteinemia as a cause of cardiovascular mortality in these subjects. Competing interests: None declared