This is among the first studies to evaluate the relationship between pathogen burden based on conventional cut-offs for seropositivity and high antibody response as potential correlates of multiple markers of inflammation. We found that high antibody response was a more consistent marker of inflammatory outcomes than seropositivity alone. Previous work has generally not differentiated between seropositivity and antibody response [29
], but our results suggest that this may be an important distinction. Earlier studies showed similar associations with inflammation in many of the pathogens examined here, suggesting that our results are compatible with population based estimates [8
]. Also in line with previous findings, sociodemographic and health indices were significantly associated with number of seropositive pathogens [28
]. There are less data, however, showing the relationships between sociodemographic characteristics and antibody response in population based studies in the US.
Testing of persistent pathogens often relies on enzyme linked immunosorbent assays (ELISA). These assays provide a qualitative assessment of the amount of antibody in a serum sample. Clinical studies are generally concerned with whether or not individuals tests positive for IgM or IgG to the pathogen of interest and rarely provide data on the qualitative antibody response identified in the ELISA test. Our findings regarding associations of antibody response with inflammation suggests that the qualitative data obtained from an ELISA is an important variable given that this measure predicts inflammation levels. Our findings suggest that antibody levels among the infected, rather than seropositivity is a better marker of inflammatory levels.
Pathogen burden (number of seropositive pathogens) was more strongly associated with inflammation than was any one individual pathogen and seropositivity to pathogens was highly correlated. Pathogen burden has been implicated in cardiovascular disease in several different study populations [33
]. It has been suggested that pathogen burden may not only be a predictor of coronary complications, as previously described, but that it may also be associated directly with the development of atherosclerotic plaques [24
]. Zhu and colleagues have suggested that inflammation is one pathway by which these associations are linked [8
]. Prasad et al
. found that pathogen burden and CRP levels interacted significantly such that carriers of 4-5 pathogens and elevated CRP levels (> 0.5 mg/dL) had higher odds of coronary artery disease compared to those with fewer pathogens [4
], supporting the notion that pathogen burden along with inflammation has a stronger impact than any one pathogen alone or inflammation alone. A recent study showed that antibody levels to one of the pathogens we examined here, CMV, was associated with an increased rate of cardiovascular mortality among aging individuals [37
]. No previous studies have examined the impact of high antibody response to multiple pathogens on inflammatory outcomes.
We found that associations of antibody levels with IL-6 were more consistent than associations with the other inflammatory markers. In addition, associations with IL-6 persisted after adjustment for age, race and socioeconomic factors but were reduced and no longer statistically significant after adjustment for potential mediators of inflammation such as smoking and obesity. Limited sample size, as well as strong associations between antibody levels and variables such as BMI and smoking may have hampered our ability to detect associations after multiple adjustments. Larger studies are needed to determine whether associations are independent of other risk factors.
There are several possible pathways by which high antibody response to multiple pathogens could impact inflammation more strongly than seropositivity alone. Infection with a particular pathogen may not strongly influence a disease process whereas a strong immune response to that pathogen may. Rheumatic heart disease is an example of this phenomenon whereby the immune response to the pathogen (and ensuing inflammation) is the relevant disease process, rather than the infection itself [38
]. In the case of infections occurring early in life and persisting over time (likes several of the pathogens we studied); a higher initial infectious dose may lead to a more vigorous antibody response at a young age which is maintained over the life course for persistent infections. Since persistent infections are generally acquired early in life [39
], the long term effects of high antibody response may lead to chronically higher levels of inflammation. However, it is not clear why individuals exhibit differentials in antibody response to pathogens. Recent work has indicated that lower socioeconomic position and exposure to chronic stressors may lead to a higher antibody response to CMV, HSV, and H. pylori
]. Indeed, there are several experimental studies that have demonstrated significant relationships between exposure to psychosocial stressors and a high antibody response to persistent herpes virus (see Herbert and Cohen 1993 for review). Thus, a number of host characteristics may influence antibody response to an infection. Additional studies are needed to better understand the mechanisms by which antibody response may influence systemic inflammation over the life course.
Some methodological considerations of this study warrant mention. Relatively small sample sizes may have affected our ability to detect associations with all markers. The inflammatory outcomes we tested could have been influenced by a number of factors that were not measured or measured imprecisely, even though care was taken to account for several potentially inflammatory conditions and medications that may impact inflammation. Our models also did not account for recent or current acute inflammatory states, such as those associated with the common cold, which may have impacted our findings. Antibody and biomarker levels were only measured once and intra-individual variation cannot be accounted for. However, assay variability would be expected to bias findings towards the null so the observed associations are potentially underestimations. Also, IgG antibody response provides data on past infection- we did not examine data on recent infection or DNA shedding of the viruses and cannot assess whether individuals were experiencing reactivation of the infection. The cross-sectional nature of this analysis precludes implications of causality, and specifically, we cannot rule out the possibility that higher inflammatory status may have led to higher levels of antibodies via activation of the immune response. Four racial/ethnic groups were represented in our data, adding an important strength to our study; however, findings may not be representative of other ethnic or age groups. Finally, it is important to note that viral and bacterial infectious agents may act through different mechanistic pathways to impact inflammation, but assessment of alternative biological pathways, such as infection residing within the cardiovasculature, was beyond the scope of this cross-sectional observational study.