Preventing Myocardial Infarction With Vaccination: Title and subTitle BreakMyths and Realities

Many clinical studies supported by basic experiments suggest that infections may trigger an acute coronary syndrome. The role of respiratory infections, especially influenza and pneumococcal pneumonia, in acute coronary syndrome in high-risk patients has been well documented.^{1 - 3} Such infections can cause an exaggerated inflammatory response in a high-risk atherosclerotic plaque, which may be followed by destabilization of the plaque, activation of the coagulation cascade, vascular thrombosis, and subsequent myocardial infarction (MI). Given the high prevalence of coronary heart disease and its risk factors in the population and the frequency of influenza and pneumonia, it is likely that each year thousands of patients might develop cardiovascular events after having such infections. This potentially causal effect has immense clinical implications, because it offers a potential method for preventing cardiovascular events by preventing or treating these infections.⁴

In this issue of JAMA, Tseng and colleagues⁵ have addressed an interesting question. Is vaccination against pneumococcal pneumonia associated with reduced risk of acute MI or stroke? In a retrospective study on a large cohort of patients enrolled in a health plan, the authors evaluated the incidence of MI or stroke in patients who had received pneumococcal vaccine vs patients who were not vaccinated. The researchers reviewed the medical records of approximately 360 000 adult men, aged 45 to 69 years, with no previous history of MI or stroke, who were followed up for an average of 4 years. After adjusting for multiple known confounders, the authors concluded that there was no association between previous pneumococcal vaccination and risk of acute MI or stroke.

One strength of this study is that it focused on a large population in a single health care system. However, several important limitations apply. Although the optimal method for addressing the issue in question would be to perform a randomized clinical trial, such a trial probably could not be conducted, because the majority of patients with coronary heart disease or its risk factors would have 1 or more indications for receiving the pneumococcal vaccine and, thus, could not ethically be randomized to receive placebo. However, the retrospective design is fraught with methodological problems. For instance, there may be substantial differences between patients who receive a vaccine and those who do not. Two main factors could influence this difference—the physician's preference to administer or withhold the vaccine (ie, physicians' adherence to guidelines) and the patient's personal decision to request, accept, or refuse the vaccine (health behaviors). The physician is more likely to recommend vaccination for patients who have more numbers of risk factors for MI or stroke than for those who do not. This bias may be offset, in part, by health-conscious patients who request the vaccine when the physician might not have thought of suggesting vaccination.⁶

Comparison of patients who are vaccinated and not vaccinated included in the study by Tseng et al⁵ reveals many substantial differences between the 2 groups. The patients who were vaccinated were older, more likely to be smokers, and more likely to have diabetes mellitus, hypertension, high cholesterol levels, and heart failure or other cardiovascular conditions. Although multivariable adjustment for potential confounders, along with additional adjustment for the propensity score, as performed by the authors, can help to account for differences in potential confounders, these adjustments are unlikely to remove the influence of such prominent discrepancies altogether. Additional concerns in the study by Tseng et al⁵ include the use of billing information instead of an actual chart review for determining vaccination exposure and clinical outcome, as well as the absence of women from the study.

The authors suggest that a possible explanation for their findings is that pneumococcal vaccine is not effective. Although controversy about vaccine efficacy persists,⁷ numerous other studies have shown that the vaccine is associated with a reduction in the risk of invasive pneumococcal infection by 30% to 60%,^{8 - 9} as well as providing significant protection against severe community-acquired pneumonia.¹⁰

The findings of Tseng et al contrast with those reported by Lamontagne et al,¹¹ who also performed a retrospective study based on billing data but reported a decrease in MI risk (nearly by half) among recipients of pneumococcal vaccine. It is difficult to determine whether these differences are due to unadjusted confounders, uncaptured biases, differences in design, low accuracy of billing databases, or simply chance.

Care should be taken in evaluating the potential effect of a vaccine for specific outcomes, such as cardiovascular events. In vaccine studies that use a case-control design, the morbidity and mortality benefits of vaccination are likely to be overestimated. The attributable risk might provide a more realistic estimate of the effect size than the odds ratio. When clinical trials are not feasible, observational studies that examine treatment outcomes should ensure careful selection of case and control groups, with proper matching, use of validated clinical data rather than billing data, use of a prospective design, stratification of study participants, and statistical adjustments for potential confounders and for propensity scores.

Another important issue involves the mechanism by which vaccination might prevent MI. Infections, especially influenza and pneumococcal pneumonia, are known to trigger acute coronary syndromes, and the association between influenza and bacterial pneumonia meets the majority of Hill's criteria for evaluating a cause-effect relationship.^{3 - 4} Therefore, a clinically effective vaccine can potentially reduce the risk of MI by preventing these triggering events. Several studies have shown that influenza vaccine is associated with reduced risk of cardiovascular events and more recent data suggest that use of antivirals after influenza may be associated with reduced risk of subsequent MI and stroke.^{4 ,12 - 13} Studies in mice have demonstrated that influenza can induce a severe systemic and local (at the plaque level) inflammatory reaction similar to the vascular inflammation observed in autopsies of persons who had sudden cardiac death. Such studies are not yet available with regard to pneumococcal pneumonia.

Future preclinical and clinical studies should evaluate whether early use of antibiotics, use of immunomodulating agents including statins, and use of antiplatelet agents such as clopidogrel and aspirin might help reduce the risk of infection-related cardiovascular events. Other studies should include exploring development of atherosclerosis vaccines based on antigens involved in atherogenesis, such as those developed against low-density lipoprotein cholesterol epitopes.¹⁴

The study by Tseng et al⁵ is another important reminder of the complexities involved in evaluating observational studies examining the relationship between vaccine administration and clinical outcomes, and the need for more robust study designs. Until rigorous data from clinical trials are available to determine whether pneumonia vaccine can prevent MI, physicians should strictly adhere to available guidelines for optimizing vaccination rates in recommended target groups, because these rates are still far from optimal.