Biomarkers of inflammation and malnutrition associated with early death in healthy elderly people

Abstract

Objectives: To determine whether malnutrition and biomarkers of inflammation predict all-cause, cancer, and cardiovascular mortality in healthy elderly subjects.

Design: Population-based study, prospective cohort.

Setting: Sète, on the French Mediterranean coast.

Participants: Five hundred and fifty-three men and 888 women aged 60 and older from the Pathologies Oculaires Liées à l'Age cohort free of known comorbidities.

Measurements: Plasma levels of cholesterol, albumin, transthyretin (TTR), C-reactive protein (CRP), and alpha 1-acid glycoprotein (AAG) were measured at baseline. To investigate the risks of 5-year (early) and 5- to 9-year (late) mortality, hazard ratios (HR) were evaluated using Cox models.

Results: In men, early death was associated with high CRP and AAG and low albumin and TTR. In women, early death was associated with high AAG, low TTR and low cholesterol. For late death, the only significant association was with CRP in men. A synergistic effect was observed between biomarkers of inflammation and malnutrition. In men, the adjusted HR of early death was 4.98 (95% confidence interval (CI)=2.25-11.01) for both CRP in the highest quartile and albumin in the lowest. In men, this risk was greatest for both AAG in the highest quartile and TTR in the lowest (HR=6.86, 95% CI=3.20-14.71). In women, this risk was greatest for both AAG in the highest quartile and TTR in the lowest (HR=4.64, 95% CI=1.79-12.05). Cancer mortality was greater for high CRP and AAG and for low albumin and TTR in men but not in women.

Conclusion: Biomarkers of inflammation and malnutrition together predict early mortality. In healthy elderly subjects, TTR and AAG could be helpful in identifying elderly subjects at higher risk of death.

Figures

Figure 1

Adjusted Hazard Ratios of early death (≤5 years) in men (●, black bar) and women (◆, grey bar) according to combined criteria. Vertical bars show 95% confidence intervals (CI). The association is significant when the CI does not contain the value one (horizontal axis).