total cholesterol to the overall predictive value of the model. For example, considering the hazard ratios reported for the three continuous variables included in the analysis, 10% relative increases in age, systolic blood pressure, and total cholesterol in women would correspond to relative increases in the cardiovascular risk of 42·3%, 19·2%, and 5·6%, respectively. Accordingly, the prognostic value of total cholesterol for the prediction of cardiovascular events found in this study was lower than previously reported in analyses from various regions of the world, including the USA.2–4 In particular, on the basis of the results of this study, an increase in total cholesterol of 1 mmol/L (38·7 mg/dL) in the range 174–271 mg/dL (mean±SD of the study population) would be associated with a relative risk increase of about 10% in women and 13% in men. Considering the average age of the study population (48 years), these values are four to ﬁve times lower than that found in a large meta-analysis5 of 61 prospective studies (1 mmol/L lower total cholesterol associated with about 50% lower risk in both sexes in middle-aged individuals). We declare that we have no conﬂict of interest.
*Piercarlo Ballo, Arianna Bocelli, Sergio Mondillo [email protected]
UO Cardiologia, Ospedale “S Andrea”, Via Veneto 197, 19100 La Spezia, Italy (PB); Meyer Hospital, University of Florence, Florence, Italy (AB); and Department of Cardiovascular Diseases, University of Siena, Siena, Italy (SM) 1
Gaziano TA, Young CR, Fitzmaurice G, Atwood S, Gaziano JM. Laboratory-based versus non-laboratory-based method for assessment of cardiovascular disease risk: the NHANES I Follow-up Study cohort. Lancet 2008; 371: 923–31. Bittner V, Hardison R, Kelsey SF, Weiner BH, Jacobs AK, Sopko G. Non-high-density lipoprotein cholesterol levels predict ﬁve-year outcome in the Bypass Angioplasty Revascularization Investigation (BARI). Circulation 2002; 106: 2537–42. Hoogeveen RC, Gambhir JK, Gambhir DS, et al. Evaluation of Lp[a] and other independent risk factors for CHD in Asian Indians and their USA counterparts. J Lipid Res 2001; 42: 631–38.
Froom P, Kristal-Boneh E, Melamed S, Harari G, Benbassat J, Ribak J. Serum total cholesterol and cardiovascular mortality in Israeli males: the CORDIS Study. Isr Med Assoc J 2000; 2: 668–71. Prospective Studies Collaboration. Blood cholesterol and vascular mortality by age, sex, and blood pressure: a meta-analysis of individual data from 61 prospective studies with 55 000 vascular deaths. Lancet 2007; 370: 1829–39.
Authors’ reply Each set of correspondents presents valid independent points of discussion, yet together they illustrate the competing challenges of risk prediction tools for cardiovascular disease that we sought to address. On one hand, important risk factors need to be included to ensure that the tool has the discriminatory power to predict major cardiovascular events and is well calibrated. On the other, the tool must be relatively simple to implement and minimise costs if it is to be adopted. Therefore, we are not surprised that Josef Niebauer and David Niederseer suggest we left out an important variable (physical activity), whereas Pascal Bovet and Fred Paccaud suggest we included one too many (diabetes mellitus), and Piercarlo Ballo and colleagues suggest that the total cholesterol variable was not as robust as in other studies. We agree with Niebauer and Niederseer that physical activity is an important independent risk factor for cardiovascular events. However, separating its eﬀects from other risk factors makes it more diﬃcult to quantify. The Framingham authors1 state that “physical activity and obesity are not included [in their model] because these factors work to a large extent through the major risk factors, and their unique contribution to CHD prediction can be diﬃcult to quantify.” Furthermore, to date we do not have a generally acceptable tool for ascertaining an accurate measurement of physical activity that has been shown to be both precise and valid in multiple cultures. We would certainly support the use of a model that included a reliable measurement of
physical activity that added signiﬁcant predictive discrimination without signiﬁcant cost. Certainly, though, the use of formal exercise testing or global positioning system devices at a cost of US$450 per device is not likely to meet this condition. We recognise that there are some countries where there might be limited information on diabetes status, particularly in rural areas. A look at how a model would perform without diabetes information would be worth investigating in individual countries where testing is extremely limited, as Bovet and Paccaud suggest. However, we would not favour the removal of diabetes history from the model for most countries, especially middleincome countries such as Mexico that do have the means for assessing diabetes. We are only recommending the opportunistic use of diabetes status in our model, and not to use this seems to be wasting valuable information. We agree with Ballo and colleagues that the risk associated with cholesterol seems to be less than that seen in other cohorts. But the risk reduction they refer to of 50% per mmol/L refers only to coronary heart disease.2 Our endpoint includes stroke, which was not found to have an association with total cholesterol in the same study, and congestive heart failure. Thus we would expect the overall association to be lower. Further, cholesterol is colinear with many of the other risk factors in the model, which might aﬀect its overall eﬀect in diﬀerent populations. Finally, we have looked at this issue brieﬂy in other cohorts and found that, even with a higher association of risk, the added value of cholesterol information to the other risk factors is negligible when it is the ﬁfth or sixth variable introduced into the model. We declare that we have no conﬂict of interest.
*Thomas A Gaziano, J Michael Gaziano [email protected]
Division of Cardiovascular Medicine, Brigham & Women’s Hospital, Boston, MA 02115, USA
www.thelancet.com Vol 371 June 21, 2008
Anderson KM, Odell PM, Wilson PW, Kannel WB. Cardiovascular disease risk proﬁles. Am Heart J 1991; 121: 293–98. Lewington S, Whitlock G, Clarke R, et al. Blood cholesterol and vascular mortality by age, sex, and blood pressure: a meta-analysis of individual data from 61 prospective studies with 55 000 vascular deaths. Lancet 2007; 370: 1829–39.
Non-speciﬁc immunomodulation in chronic heart failure We read the results of the ACCLAIM trial (Jan 19, p 228)1 and, although the trial failed to meet its primary endpoints, we are encouraged that researchers are exploring the immune system in congestive heart failure. However, we have reservations about the method used in this trial and believe the conclusions (based on subanalysis data) could prove erroneous. The concept that a small volume of autologous, apoptotic leucocytes injected into the gluteus maximus would systemically skew the immune response from inﬂammatory to anti-inﬂammatory seems unlikely. Guillermo Torre-Amione and colleagues base their theory on in-vitro work reporting that the immune system secretes anti-inﬂammatory cytokines after interaction with apoptotic cells. Yet apoptosis occurs 50–70 billion times per day without immunomodulatory eﬀect. Moreover, apoptosis acts as a regulatory mechanism during propriocidal regulation,2 via induction of tumour necrosis factor α,3 or during passive withdrawal.4 These mechanisms are not associated with skewed anti-inﬂammatory responses. Furthermore, induced stress causes abnormal cell death rather than programmed cell death. This abnormal cell death results in stress protein expression and intracellular protein release, leading to inﬂammatory (rather than anti-inﬂammatory) pathways.5 These issues are applicable to the ACCLAIM study, where cell stress www.thelancet.com Vol 371 June 21, 2008
was rapidly induced via ultraviolet light and oxidative stress. Research into immunomodulatory treatments for congestive heart failure has been stimulated after reports of elevated cytokine concentrations in patients with this disorder. This discovery has led researchers to attempt to control cytokine activity, rather than to understand the speciﬁc immunological mechanisms involved within the disease. We believe a great deal more research is required before successful immunotherapy for CHF is realistically achieved. We declare that we have no conﬂict of interest.
*James E Fildes, Steven M Shaw, Nizar Yonan, Simon G Williams james.ﬁ[email protected]
University Hospital of South Manchester NHS Foundation Trust, Manchester M23 9LT, UK 1
Torre-Amione G, Anker SD, Bourge RC, et al. Results of a non-speciﬁc immunomodulation therapy in chronic heart failure (ACCLAIM trial): a placebo-controlled randomised trial. Lancet 2008; 371: 228–36. Boehme SA, Lenardo MJ. Propriocidal apoptosis of mature T lymphocytes occurs at S phase of the cell cycle. Eur J Immunol 1993; 23: 1552–60. Zheng L, Fisher G, Miller RE, et al. Induction of apoptosis in mature T cells by tumour necrosis factor. Nature 1995; 377: 348–51. Jones LA, Chin LT, Longo DL, et al. Peripheral clonal elimination of functional T cells (1990) Science 1990; 250: 1726–29. Matzinger P. Tolerance, danger, and the extended family. Annu Rev Immunol 1994; 12: 991–1045.
previously halted phase III Simpadico trial.4 Proponents of Celacade favour the concept that, after intramuscular administration of denatured blood, immune modulation ensues, with upregulation in the production of anti-inﬂammatory cytokines such as interleukin 10 and transforming growth factor β. Because chronic heart failure and limb ischaemia are disorders linked to inﬂammation and chronic oxidative stress, the reduction of proinﬂammatory cytokines is assumed to downregulate chronic inﬂammation and delay the progress of disease. I dissent from this interpretation because the biological and clinical eﬀects of judiciously small ozone doses in vasculopathies are based on other crucial phenomena such as vasodilatation, increased delivery of oxygen in ischaemic tissues, and upregulation of antioxidant enzymes, especially haem oxygenase I.5 Immunomodulation might be only an additional factor. The use of a masked saline placebo is also objectionable: it would have been more appropriate to use oxygenated autologous blood.
We declare that we have no conﬂict of interest.
Velio Bocci [email protected]
University of Siena, 53100 Siena, Italy
The immunomodulatory treatment used in Guillermo Torre-Amione and colleagues study1 (the Celacade system) is an expensive approach based on intramuscular injection of a 10 mL autologous blood sample treated with an excessive dose of ozone, ultraviolet light, and heat (42·5C) for 20 min and then given by intragluteal injections at several points during no fewer than 22 weeks. The extremely high oxidation of blood with ozone and ultraviolet light, and the additional heating, causes a denaturation of blood that contrasts with the practice of ozonated minor autohaemotherapy2 and has proved useless in an AIDS trial3 and in the
Torre-Amione G, Anker SD, Bourge RC, et al. Results of a non-speciﬁc immunomodulation therapy in chronic heart failure (ACCLAIM trial): a placebo-controlled randomised trial. Lancet 2008; 371: 228–36. Bocci V. The case for oxygen-ozone therapy. Bxci 2007; 64: 44–49. Garber GE, Cameron DW, Hawley-Foss N, et al. The use of ozone-treated blood in the therapy of HIV infection and immune disease: a pilot study of safety and eﬃcacy. AIDS 1991; 5: 981–84. Olin JW, Hiatt WR, Mohler E, et al. A multicenter, randomized, double-blind, placebo-controlled study of immune modulation therapy in patients with symptomatic peripheral arterial disease: the SIMPADICO Trial. Presented at the American College of Cardiology 55th Annual Scientiﬁc Session; March 11–14, 2006; Atlanta, GA, USA. Bocci V, Aldinucci C, Mosci F, et al. Ozonation of human blood induces a remarkable upregulation of heme oxygenase-1 and heat stress protein-70. Mediators Inﬂamm 2007; 2007: 26785.