In the current prospective cohort study of the U.S. population with diabetes, we found that aging-related makers were associated with mortality. Specifically, Phenotypic and Biological Age showed positive associations with all-cause, CVD, and cancer mortality, higher telomere length was associated with lower all-cause mortality, and Klotho had a U-shaped relationship with all-cause, CVD, and cancer mortality. Further, metformin decreased all-cause, CVD, and cancer mortality risk. More importantly, such associations were partly mediated by Phenotypic Age and Biological Age. These findings suggested that the associations of aging with mortality were significant, and anti-aging was an important approach to reduce death among diabetics.
Previously, a meta-analysis of 17 cohorts with 5575 diabetics and 6439 control subjects showed shorter telomere length in diabetes (22). An inverse association was noted between telomere length and mortality among Chinese, Danish, and Italian diabetics (HR=1.87-3.45) (23). Consistent findings were also pronounced in our analysis among the U.S. diabetics. Telomere length is accepted as an indicator of cellular aging. Interestingly, removing aging cells or delaying their formation in mice can improve diabetes progress and its complications (10, 24). However, the aging process of mouse cardiomyocytes and human lungs is only linked to the telomere-associated DNA damage, without significant telomere shortening detected (14, 15). The above results may be attributed to the limited role of telomere in interpretation at organs or the whole body level. Importantly, diabetics often develop multiple organ disorders and die by complications. Hence, more comprehensive indicators which can represent the body’s status were needed.
Phenotypic and Biological Age incorporated composite clinical biomarkers, such as inflammation (e.g. CRP), immunity (e.g. number of immune cells), and organ function (e.g. albumin and serum urea nitrogen) (25). They are useful to identify the unhealthy condition of patients caused by multiple organ disorders. Herein, we found that Phenotypic Age and Biological Age were significantly associated with all-cause and cause-specific mortality among diabetics. Moreover, we defined premature aging status based on chronological age and found that diabetics who appeared older than expected physiologically had a higher death risk. The previous study also reported similar associations of Phenotypic and Biological Age with all-cause mortality, but in the general population (HR=1.09, 1.10, respectively). Notably, among the disease-specific mortality, the risk of diabetes-caused mortality increased mostly (HR=1.20) (16). These researches indicated the importance of the whole body’s aging for death among diabetics.
On the other hand, as the aging-related molecule, Klotho has been reported to be involved in the aging process for over 30 years, where it regulates phosphate homeostasis, insulin, and Wnt signaling (26). Interestingly, in our study population, Klotho concentration had a U-shaped relationship with all-cause, CVD, and cancer mortality. Previous epidemiological studies in the old or chronic kidney disease population revealed that Klotho concentration was negatively associated with death (27, 28). Indeed, overexpression of soluble Klotho up-regulated fibroblast growth factors-23 levels, which may lead to hypovitaminosis D (29). More importantly, a recent prospective cohort study including 6,329 diabetics reported that the level of serum 25-Hydroxyvitamin D was negatively associated with all-cause and cause-specific mortality (30). These studies supported our findings that circulating Klotho may have a U-shaped relationship with mortality risk in diabetics. Collectively, from many perspectives, we observed the effect of aging on promoting diabetics’ death.
Metformin was first introduced to the world in 1957, as an anti-hyperglycemic agent. Consistent with previous studies, we found that metformin users had a lower risk of all-cause and cause-specific mortality, which were positively correlated with the number of days taken among diabetics. Recently, accumulating evidence has revealed the gerotherapeutic effect of metformin on lowering the incidence of multiple age-related diseases and all-cause mortality in diabetics (31). A meta-analysis indicated that the anti-aging function of metformin is possibly independent of its effect on diabetes control (32). Hence, we further investigated whether anti-aging was involved in the role of metformin in preventing mortality among diabetics. We found that metformin use decreased mortality risk among diabetics, regardless of whether HbA1C was well controlled. Of note, our mediation analyses showed that metformin reduced mortality partly by decreasing Biological Age and Phenotypic Age, emphasizing the importance of anti-aging. Mechanism-related studies in multiple models have elucidated metformin participated in various pathways of aging, including deregulated nutrient sensing, altered intercellular communication, genomic instability, and loss of proteostasis (33, 34). Therefore, treatments targeting anti-aging are greatly promising in improving the mortality of diabetics. Some interventions have been well validated in animal or cells models (35). For example, senolytic drugs (e.g. ABT263) improved glucose tolerance and insulin sensitivity in diabetes mice by reducing senescent cell burden (10). Another combination senolytic agents, dasatinib and quercetin (DQ), offered a good therapeutic effect on mice with age-related diseases (36). In particular, preliminary clinical trials presented the positive effects of DQ on patients with idiopathic pulmonary fibrosis and diabetic kidney disease (9, 37). The current study provided evidence that anti-aging was effective to delay the progression of diabetes and mortality.
This study has some strengths. Firstly, the study was a prospective cohort study based on data from a large nationally representative survey among U.S. diabetics. Secondly, the associations reported in this study were relatively robust by adjustment for a variety of confounders and several sensitivity analyses. Finally, we examined the relationship between aging and death outcomes among diabetics from multiple perspectives. There are also limitations. Firstly, considering the cross-sectional nature of the NHANES data, the time-varying changes in aging markers could not be investigated. Secondly, the study lacked further information on the severity of diabetes, though adjusting for diabetes medication use, duration of diabetes, HbA1c levels, and some self-reported comorbidities. Thirdly, unmeasured confounders and measurement errors may bias our analyses. Finally, whether anti-aging is involved in the effect of metformin on mortality still needs further research. In addition, we only analyzed the role of one anti-aging drug in mortality, which limited the interpretation of our conclusions.