Biomarker-disease relationships are extensively investigated. However, associations between common clinical biomarkers and healthspan, the disease-free lifespan, are largely unknown. We aimed to explore the predictive values of ten biomarkers on healthspan and lifespan, and to identify putative causal mechanisms.
Using data from 12,098 Swedish individuals aged 47-94 years, we examined both serum concentrations and genetically predicted levels of ten glycemic, lipid-, inflammatory, and hematological biomarkers. During a follow-up period of up to 16 years, 3681 incident cases of any chronic disease (i.e., end of healthspan) and 2674 deaths (i.e., end of lifespan) were documented. Cox regression models were applied to estimate the associations of a one standard deviation increase in biomarkers with healthspan and lifespan.
Scientists observed that higher circulating levels of HbA1c, FBG, CRP, and TG were indicative of a higher risk of healthspan ending; in contrast, increased levels of HDL-C, ApoA1, TC were associated with a lower risk of any chronic disease; no statistically significant evidence of association was observed after multiple testing correction for LDL-C, Hb and ApoB. Of all healthspan-detrimental biomarkers, glycemic biomarkers exhibited the largest effect sizes, with a one-SD increase in HbA1c associated with a 29% increased risk (HR 1·29, 95% CI 1·24–1·34]). Within all healthspan-beneficial biomarkers, those related to HDL metabolism showed the greatest effects, in which a one-SD increase in HDL-C was associated with a 8% decreased risk (HR 0·92, 95% CI 0·89–0·96). In an independent Swedish cohort, SATSA, we also observed statistically significant results for serum FBG, CRP, and HDL-C in association with heathspan, as well as directionally consistent evidence for serum HbA1c, TC, and ApoA1. With regard to lifespan associations, we found all clinical biomarkers were statistically significantly associated with death; specifically, Hb, ApoB, and LDL-C were inversely associated with death risks, while the other biomarkers showed similar patterns as with healthspan.
In the multiple-biomarker models, the two sets of biomarkers selected were only slightly different. Specifically, FBG, HbA1c, TG, TC, and CRP provided independent information associated with both healthspan and lifespan.
Scientists observed a statistically significant relationship between FBG PRS and the risk of any chronic disease with an HR of 1·05 (1·02, 1·09), meaning a one-SD increase in genetic predisposition to elevated blood glucose level was associated with a 5% higher risk (HR 1·05, 95% CI 1·02–1·09). The other glycemic PRS, HbA1c PRS, showed a weak association in the same direction, but was not statistically significant. Results for other clinical biomarkers PRSs showed either null or very weak associations
Additional models showed that the effect of FBG PRS on any chronic disease was independent of other non-glycemic markers, including BMI, as well as serum TG, HDL-C, LDL-C, CRP, and Hb (HR 1·05, 95% CI 1·01–1·09). However, controlling for serum FBG led to a substantial attenuation in effect estimate (HR 1·01, 95% CI 0·98–1·05). The result suggests FBG PRS-healthspan association could be explained by serum FBG level to a large extent.
Lifespan associations showed statistically significant evidence suggesting genetically predicted higher CRP was associated with a lower risk of death, with an HR of 0·91 (95% CI 0·87–0·95) estimated for a one-SD increase in CRP PRS. HR estimates also suggested an increase in the PRS predisposing to higher circulating levels of TC, LDL-C, and ApoB increased the risk of death, albeit with statistically non-significant P values after multiple testing correction. No robust evidence of lifespan association was observed for the other PRSs.
Further, Scientists found neither serum CRP nor other markers (BMI and serum TG, LDL-C, HDL-C, FBG, and Hb) could explain the beneficial survival effect of CRP PRS. Particularly, the results remained intact when we replaced the whole-genome CRP PRS derived from 95 SNPs with a regional CRP PRS derived from 4 SNPs located in the CRP gene region.
Generally, the directions of biomarker associations were consistent in the “leave-one-disease-out” sensitivity analysis; most associations were more pronounced (HR further from 1) after excluding cancer cases, while less pronounced (HR closer to 1) after excluding diabetes cases. Particularly, the effects of glycemic biomarkers were largely attenuated with the exclusion of diabetes. The exclusion of diseases other than cancer and diabetes barely changed the results. After exclusion of outliers, we observed slightly changed point estimates of HRs and CIs.
Circulating concentrations of glycemic, lipid-, and inflammatory biomarkers are predictive of healthspan and lifespan. Glucose control is a putative causal mechanism and a potential intervention target for healthspan maintenance.
Sherry
Phone: 
E-mail: 
Address: