Frailty mediating the causality between leucocyte telomere length and mortality: a cohort study of 440,551 UK Biobank participants

Abstract

Introduction

Previous studies reported leucocyte telomere length (LTL) and frailty were associated with mortality, but it remains unclear whether frailty serves as a mediator in the relationship between leucocyte telomere length and mortality risk. This study aimed to evaluate how measuring LTL and frailty can support early monitoring and prevention of risk of mortality from the prospective of predictive, preventive, and personalized medicine (PPPM/3PM).

Methods

We included 440,551 participants from the UK Biobank between the baseline visit (2006–2010) and November 30, 2022. The time-dependent Cox proportional hazards model was conducted to assess the association between LTL and frailty index with the risk of mortality. Furthermore, we conducted causal mediation analyses to examine the extent to which frailty mediated the association between LTL and mortality.

Results

During a median follow-up of 13.74 years, each SD increase in LTL significantly decreased the risk of all-cause [hazard ratio (HR): 0.94, 95% confidence interval (CI): 0.93–0.95] and CVD-specific mortality (HR: 0.92, 95% CI: 0.90–0.95). The SD increase in FI elevated the risk of all-cause (HR: 1.35, 95% CI: 1.34–1.36), CVD-specific (HR: 1.47, 95% CI: 1.44–1.50), and cancer-specific mortality (HR: 1.22, 95% CI: 1.20–1.24). Frailty mediated approximately 10% of the association between LTL and all-cause and CVD-specific mortality.

Conclusions

Our results indicate that frailty mediates the effect of LTL on all-cause and CVD-specific mortality. There findings might be valuable to predict, prevent, and reduce mortality through primary prevention and healthcare in context of PPPM.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13167-024-00355-7.

Introduction

Mortality reduction is essential in the context of prediction, prevention, and personalized medicine

According to the World Health Organization (WHO) statistics, in 2019, approximately 18.5 million and 10 million people died because of cardiovascular diseases (CVDs) and cancers, respectively, accounting for over 50% of all deaths worldwide [1]. Thus, they have been regarded as important health problems and major contributors to the global burden of disease. Burden of disease refers to the health, economic, and social impact caused by diseases on individuals, communities, or entire societies, integrating factors like incidence, mortality, disability, the reduction in quality of life, and so on. The global population is ageing rapidly, resulting in an increasing burden on the health and economic systems [2]. The most two leading contributors to disease burden in older people are cardiovascular diseases (30.3% of the total burden in people aged 60 years and older) and malignant neoplasms (15.1%) [3]. The pathogenesis of them has been explored in a number of studies, but it is not yet known clearly. Novel biomarkers and easily modifiable targets of prediction, prevention, and intervention should be sought. Predictive, preventive, and personalized medicine (PPPM/3PM) is an integrated strategy in healthcare that enables the recognition of individual susceptibility to disease and offers targeted preventive interventions and treatments [4–6]. Previous evidence has demonstrated that there is significant potential for reducing mortality through PPPM [7–9]. Therefore, identifying effective strategies for identifying individuals at high risk of mortality, providing primary care to decrease mortality, and delivering personalized medical interventions that ultimately reach the goals of predicting, preventing, and reducing mortality, extending life expectancy, and improving quality of life are paramount.

Shorter telomere is associated with elevated risk of mortality

Telomeres consist of repetitive DNA sequences, TTAGGG, located at the end of linear chromosomes and their primary function is to protect chromosomes from damage [10]. During cell division, telomeres shorten due to oxidative stress and the end replication problem [11]. When critically shortened, it can result in cell cycle arrest, cell apoptosis, or senescence [12]. Previous literature reports that telomerase [13] and oxidative stress [14] were critical factors of LTL and are difficult to change through medical interventions from primary care. Some epidemiological researches have reported that shorter leucocyte telomere length (LTL) is associated with many age‐related diseases, such as CVDs [15, 16], diabetes [17, 18], and cancers [19, 20]. Furthermore, a previous study reported that a shorter LTL was associated with an elevated risk of all-cause and CVD-specific mortality, but there was no association with cancer-specific mortality [21]. It is essential to identify modifiable factors within their association to implement interventions aimed at reducing mortality rates.

Frailty is associated with mortality risk and is a potential interventional target for PPPM

Over the past few decades, frailty has received increasing attention [22]. Frailty is defined as an age-associated state characterized by a decline in physiological reserve and loss of resistance to stressors resulting from accumulated age-related deficits [23]. Fried’s frailty phenotype (FP) and frailty index (FI) models are the two main instruments used to measure frailty [24]. Frailty has been observed to be associated with negative health outcomes, such as hospitalizations [25] and mortality [26]. Additionally, frailty was a modifiable risk factor, which could be refined by interventions of PPPM. To our knowledge, no specific studies have assessed the mediating role of frailty in the association of LTL with mortality. It remains unclear whether LTL is directly associated with mortality risk or whether frailty partly mediates this association. It is vital to identify additional potential biomarkers or modifiable factors in the context of PPPM to achieve the purpose of predicting, preventing, and early intervention in reducing mortality risk.

Working hypothesis in the framework of PPPM

In this large prospective cohort study of the UK Biobank, we aimed to examine the effect of LTL and frailty on all-cause and cause-specific mortality and assess the effect of LTL and frailty as predictive biomarkers of mortality in the frame of PPPM. Additionally, we targeted to assess the extent to which the association between LTL and mortality is mediated by frailty. Previous researches revealed the possibility of decreased risk of CVDs [9, 27], cancer [28, 29], and mortality [8] in the framework of PPPM. From the perspective of PPPM, if frailty mediates the association between LTL and mortality, there would be helpful to early identify frailty status, early screen high-risk population, timely predict and prevent the risk of mortality, and thus extend life expectancy.

Methods

Study participants

UK Biobank is a prospective population-based cohort study enrolling approximately 500,000 participants aged 40–69 years in 2006 to 2010 across England, Scotland, and Wales [30]. More details of the UK Biobank can be found elsewhere online: http://www.ukbiobank.ac.uk. Ethical approval of the UK Biobank was obtained from the National Health Service National Research Ethics Service, and all participants provided written informed consent. We report this study based on the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement [31]. The study was carried out under the UK Biobank project number 95259.

Participants who had missing data with LTL, underwent bone marrow transplantation before the baseline examination, or had over 20% missing data with the 49 items for calculating the frailty index (n = 80,127) were excluded. We further excluded participants who had accidental deaths, had missing covariate data, or died within 2 years from baseline (n = 11,672). Consequently, 440,551 individuals were included in the main analysis of the association between LTL and all-cause mortality (Fig. 1).

Fig. 1.

Flow chart of the study

Exposures

LTL, extracted from peripheral blood leukocytes and measured using quantitative PCR, was a ratio of telomere repeat copy number to the single copy gene (T/S) and then adjusted for the influence of technical parameters [32]. In this study, we use the field 22192 in UK Biobank, in which LTL has been additionally loge-transformed and then Z-standardized.

Outcomes

Follow-up time was counted from the date of the assessment centre visit until the date of death or the date of censoring (November 30, 2022), whichever came first. The primary outcome was all-cause mortality. Dates of death were provided by reviewing the death certificates held by the National Health Service Information Centre (England and Wales) and the National Health Service Central Register Scotland (Scotland). The secondary outcome was cause-specific mortality, including CVD and cancer-specific mortality, which was identified from the underlying (primary) cause of death in the death registry. We used ICD-10 codes I00-I99 to define deaths due to CVD and C00-C97 to define deaths due to cancer.

Mediator

Frailty refers to a state of increased vulnerability and reduced physiological reserve in individuals, commonly assessed using the frailty index and frailty phenotype. FI, a mediator used in the primary analysis, was calculated for participants on the basis of 49 self-reported deficits according to a previous study [33]. Each deficit was rescored in accordance with its severity. Then, the sum was divided by the total number of deficits to obtain FI for each participant. For better comparability and reliability, FI was standardized. Additionally, according to the literature [34], participants with a score over 0.2 were categorized as frail, and those with a score less than 0.2 were categorized as nonfrail. In the sensitivity analysis, we also considered FP [26], which was evaluated through five self-reported or objectively measured components: weakness, slowness, exhaustion, low physical activity, and unintentional weight loss [25]. FP scores ranged from 0 to 5, and higher scores suggested greater frailty. Following previous studies [25, 26, 35], participants were categorized as nonfrail (FP score ≤ 2) and frail (FP score ≥ 3).

Covariates

For this study, confounding covariates included age, sex, ethnicity, townsend deprivation index, education level, smoking status, and alcohol status. Age was defined as a continuous variable. Sex was defined as a binary variable (female vs. male). Ethnicity was classified as a binary categorical variable (white vs. other). The townsend deprivation index (TDI) was defined as a continuous variable [36]. TDI was a composite score based on the percentage of unemployment, percentage of car ownership, percentage of home ownership, and household overcrowding, with higher scores representing higher levels of deprivation [36]. Each participant is assigned a score corresponding to the output area in which their postcode is located. Education level was defined into three groups: primary or lower level, secondary level, and higher level. Smoke status was categorized into current smoker and current nonsmoker. Alcohol status was dichotomized into current drinker and current nondrinker.

Statistical analysis

Descriptive statistics were used to describe the baseline characteristics among participants in the alive and deceased groups. Continuous variables are presented as the mean ± standard deviation (SD) when normally distributed or median with interquartile range (IQR) when the variables did not follow a normal distribution, and categorical variables are expressed as frequencies and percentages. The Wilcoxon rank sum test was used for the analysis of continuous variables, and the χ² test was used for categorical variables.

We performed a generalized linear model to evaluate the effect of LTL on FI. Additionally, the time-dependent Cox proportional hazard regression models were applied to assess the association between LTL, FI, and the risk of mortality, including all-cause, CVD-specific, and cancer-specific mortality.

To examine the extent to which FI potentially mediated the association of LTL with the risk of 1) all-cause mortality, 2) CVD-specific mortality, and 3) cancer-specific mortality, mediation analyses were conducted using a counterfactual-framework approach [37] through use of the SAS PROC CAUSALMED procedure (Supplement Fig. 1). We calculated 95% CIs for total effect (TE), natural direct effect (NDE), and natural indirect effect (NIE) by the delta method [37, 38]. The NDE represented the effect of LTL on mortality that was independent of FI. The NIE indicated that the effect of LTL on mortality could be explained by its association with the mediator in the model. The proportion of the association by the mediator was evaluated to quantify the magnitude of mediation.

Overall, we built three models to calculate hazard ratios (HRs) and corresponding 95% CIs in all analyses mentioned above. Model 1 gave the unadjusted association (crude model). Model 2 adjusted for age and sex, while Model 3 additionally adjusted for ethnicity, townsend deprivation index, education level, smoking and alcohol status.

Subgroup analyses according to sex and age (< 60 and ≥ 60 years) were conducted. We also performed several sensitivity analyses to verify the robustness of the results. We repeated the primary analysis. First, to minimize any potential reverse causation, we examined the role of FI in the association between LTL and mortality among only participants with more than 4 years of follow-up (n = 437,290). Second, to remove the effect of COVID-19, the cut-off for follow-up was restricted to December 30, 2019 (n = 428,155). Then, we repeated the main analysis. Third, instead of FI, FP was used as a mediator for all analyses to test whether FI could adequately represent frailty (n = 152,346). Fourth, considering the influence of the missing data on our analyses, we conducted multiple imputation using chained equations (MICE) [39] to impute missing values of covariates and repeated the main analysis (n = 447,910). Finally, FI was used in the repeated analyses as a categorical variable.

Data were analyzed from March 2023 to August 2023. A two-sided P < 0.05 or a 95% CI for HR excluding 1.00 was considered statistically significant. All statistical analyses were conducted by the survival package in R, version 4.0.5, STATA 17.0, and SAS, version 9.4.

Results

Baseline characteristics

The baseline characteristics of the study participants are displayed in Table 1. The median (IQR) age was 58.00 (50.00, 63.00) years, and 54.39% of the participants were female. Participants with increased mortality risk were more likely to be older, male, white, have lower education attainment, with higher TDI, smoker, and nondrinker (all P < 0.001). As expected, shorter LTL and higher FI were significantly associated with the risk of mortality (all P < 0.001). During an average follow-up period of 13.74 (1.75) years, 34,002 (7.72%) death events were recorded, of which 7400 were from CVD and 17,292 were from cancer in the study.

Table 1.

Description of baseline characteristics

Characteristic	Total (N = 440,551)	Alive (n = 406,549)	Deceased (n = 34,002)	P
Age (years), median (IQR)	58.00 (50.00, 63.00)	57.00 (50.00, 63.00)	63.00 (59.00, 67.00)	< 0.001
Female, n (%)	239,608 (54.39)	225,486 (55.46)	14,122 (41.53)	< 0.001
Ethnicity, White, n (%)	419,857 (95.30)	386,845 (95.15)	33,012 (97.09)	< 0.001
Educational attainment, n (%)				< 0.001
High	146,699 (33.30)	138,593 (34.09)	8106 (23.84)
Intermediate	221,010 (50.17)	205,429 (50.53)	15,581 (45.82)
Low	72,842 (16.53)	62,527 (15.38)	10,315 (30.34)
Townsend deprivation index, median (IQR)	− 2.21 (− 3.68, 0.39)	− 2.24 (− 3.69, 0.31)	− 1.78 (− 3.46, 1.32)	< 0.001
Smoking status, n (%)				< 0.001
Current nonsmoker	395,230 (89.71)	367,558 (90.41)	27,672 (81.38)
Current smoker	45,321 (10.29)	38,991 (9.59)	6330 (18.62)
Alcohol status, n (%)				< 0.001
Current nondrinker	33,541 (7.61)	29,777 (7.32)	3,764 (11.07)
Current drinker	407,010 (92.39)	376,772 (92.68)	30,238 (88.93)
Leucocyte telomere length, median (IQR)	0.00 (− 0.64, 0.65)	0.02 (− 0.62, 0.66)	− 0.22 (− 0.88, 0.44)	< 0.001
Frailty index, median (IQR)	− 0.15 (− 0.71, 0.60)	− 0.19 (− 0.80, 0.52)	0.34 (− 0.36, 1.19)	< 0.001
Leucocyte telomere lengtha, n (%)				< 0.001
Shorter	110,136 (25.00)	98,780 (24.30)	11,356 (33.40)
Medium	220,276 (50.00)	204,046 (50.19)	16,230 (47.73)
Longer	110,139 (24.96)	103,723 (25.51)	6,416 (18.87)
Frailty, n (%)				< 0.001
Nonfrail	336,097 (76.29)	315,602 (77.63)	20,495 (60.28)
Frail	104,454 (23.71)	90,947 (22.37)	13,507 (39.72)

^aLeucocyte telomere length has been loge-transformed and then Z-standardized

The associations of leucocyte telomere length with mortality mediated by frailty

First, Table 2 shows that LTL was inversely correlated with FI (β =  − 0.021, 95% CI: − 0.024, − 0.019) in the fully adjusted model. Otherwise, the results were consistent within subgroups of sex and age (female: β =  − 0.026, 95% CI: − 0.030, − 0.022, male: β =  − 0.016, 95% CI: − 0.021, − 0.012, age < 60: β =  − 0.022, 95% CI: − 0.025, − 0.188, age ≥ 60: β =  − 0.021, 95% CI: − 0.026, − 0.017).

Table 2.

Associations of leucocyte telomere length with frailty index

Group	HR (95%CI)
	Model 1a	Model 2b	Model 3c
Total	− 0.054 (− 0.057, − 0.051) *	− 0.030 (− 0.033, − 0.027) *	− 0.021 (− 0.024, − 0.019) *
Female	− 0.062 (− 0.066, − 0.058) *	− 0.034 (− 0.038, − 0.029) *	− 0.026 (− 0.030, − 0.022) *
Male	− 0.060 (− 0.065, − 0.056) *	− 0.026 (− 0.030, − 0.021) *	− 0.016 (− 0.021, − 0.012) *
Age < 60	− 0.040 (− 0.043, − 0.035) *	− 0.031 (− 0.035, − 0.027) *	− 0.022 (− 0.025, − 0.018) *
Age ≥ 60	− 0.027 (− 0.031, − 0.022) *	− 0.028 (− 0.033, − 0.024) *	− 0.021 (− 0.026, − 0.017) *

^aUnadjusted

^bFurther adjusted for age and sex based on Model 1

^cFurther adjusted for ethnicity, townsend deprivation index, education level, smoking status, and alcohol status based on Model 2

^*P < 0.001

Second, for the effect of LTL on all-cause and cause-specific mortality, we found that each SD increase in LTL significantly decreased the risk of all-cause [hazard ratio (HR): 0.94, 95% confidence interval (CI): 0.93–0.95] and CVD-specific mortality (HR: 0.92, 95% CI: 0.90–0.97). However, LTL had no association with cancer-specific mortality (HR: 0.99, 95% CI: 0.98–0.01) after adjusting for confounders. We also found positive associations between FI and mortality (Fig. 2 and Supplement Table 1).

Fig. 2.

Forest plot for associations between leucocyte telomere length, frailty index, and mortality

The SD increase in FI elevated the risk of all-cause mortality (HR: 1.35, 95% CI: 1.34–1.36) and cause-specific mortality, including CVD-specific mortality (HR: 1.47, 95% CI: 1.44–1.50) and cancer-specific mortality (HR: 1.22, 95% CI: 1.20–1.24). Values are presented as HR (95% CIs) adjusted for age, sex, ethnicity, townsend deprivation index, education level, smoking status, and alcohol status.

Finally, Fig. 3 and Supplement Table 2 present the total, natural direct, and natural indirect associations of LTL with mortality as well as the proportion mediated. We estimated that frailty mediated 10.06% and 10.30% of the association of LTL with all-cause and CVD-specific mortality, respectively (Supplement Table 2). There was no evidence for mediation through frailty for LTL and cancer-specific mortality. Similar results were observed in stratified analyses for age (Supplement Table 3). Subgroup analyses stratified by sex found significant results for the effect of frailty on the association between LTL and cancer-specific mortality (female: − 19.63%, male: 10.44) (Supplement Table 3).

Fig. 3.

Causal mediating effect of frailty on the association between leucocyte telomere length and mortality. A All-cause mortality, B CVD-specific mortality, and C cancer-specific mortality. Values are presented as HR (95% CIs) adjusted for age, sex, ethnicity, townsend deprivation index, education level, smoking status, and alcohol status. ^*P < 0.001

Sensitivity analyses

In the sensitivity analysis, the exclusion of the participants who died prematurely within 4 years (Supplement Table 4) or after December 30, 2019 (Supplement Table 5), did not substantially change the results. Interestingly, nonsignificant results were obtained for the mediation analysis when frailty was assessed with the FP instead of the frailty index (Supplement Table 6). We did not find any evidence for mediation through FP for LTL on CVD-specific and cancer-specific mortality. We did not observe any substantial changes in the results after multiple imputation for missing values of covariates (Supplement Table 7) or considering FI as a categorical variable (Supplement Table 8).

Discussion

Summary of research findings

From the prospective of PPPM, discovering biomarkers, identifying modifiable risk factors, predicting, preventing, and performing early personalized interventions are highly important for preventing and reducing the risk of mortality. Moreover, evaluating biomarkers for obtaining predictive information on mortality risk and identifying high-risk populations early are valuable. Then, timely personalized interventions should be implemented, targeting modifiable risk factors in high-risk populations to prevent disease, mitigate the risk of mortality, and increase the life expectancy. In this large prospective study of the UK Biobank, we found that LTL was significantly associated with the risk of all-cause mortality and CVD-specific mortality. Furthermore, frailty was related to all-cause, CVD-specific, and cancer-specific mortality. Finally, it showed that frailty mediated approximately 10% of the relationships. These findings suggest LTL and frailty measuring might be useful to predict the risk of mortality, provide early screening of the high-risk population, and thus impose personalized interventions for modifiable frailty items in the context of PPPM, which is important to reduce mortality and extend life expectancy.

Comparations with previous studies

The inverse association between LTL and frailty observed in this study is not surprising, which was generally consistent with previous observational studies [40, 41]. Nevertheless, Mendelian randomization studies found no significant causal association between LTL and frailty [42, 43]. Therefore, a higher level of evidence is required to confirm this hypothesis.

In this study, we found that a shorter LTL was associated with elevated risk of all-cause and CVD-specific mortality, which was compatible with many studies conducted in different populations [21, 44, 45]. Mortality could be reduced by 6–8% through suppressing LTL shortening. Intervention for frailty could obtain great benefits and decrease mortality by 22–47%. Considering the association of LTL with cancer-specific mortality, our results are broadly in accordance with an American study [45]. However, an analysis of 3225 middle-aged women found that a shorter LTL was associated with an increased risk of cancer-specific mortality (HR = 0.85, 95% CI: 0.71–1.00) [46]. The small sample size and broad 95% CI may be interpretations for the inconsistency of results. Additionally, frailty was associated with an elevated risk of all-cause, CVD-specific, and cancer-specific mortality. Previous studies [26, 47–49] in different populations have shown similar results. HR was close to 1 in our study, possibly because of latent confounding factors. While the findings require validation in other populations, they still have some helpful implications. The authors suggested the use of LTL and frailty to predict the risk of mortality, identify high-risk populations early, and implement personalized interventions.

Notably, we demonstrated that frailty mediated approximately 10% of the association between LTL and all-cause and CVD-specific mortality. LTL shortening is recognized to serve as a key indicator of cellular senescence and organismal aging and is associated with oxidative stress and chronic inflammation [50]. Variations in oxidative stress biomarkers and chronic inflammation are frequently related to frailty status in older people [51, 52]. This could be a mechanism by which frailty mediates LTL and mortality but needs to be further investigated. The relationship between frailty and mortality could be valid across studies of different populations [26, 47–49]. The results from the UK Biobank reported that individuals with frailty have a higher risk of mortality in all age and sex subgroups [26]. The China Kadoorie Biobank showed that each 0.1 increase in the frailty index significantly elevated the risk of all-cause mortality (hazard ratio (HR) = 1.68, 95% CI 1.66–1.71), death from ischemic heart disease (HR = 1.89, 95% CI 1.83–1.94), and death from cancer (HR = 1.19, 95% CI 1.16–1.22) [47]. Although the exact underlying mechanisms are not yet known, this evidence supports our discovery that frailty partly mediates the associations of LTL with all-cause and CVD-specific mortality. From the viewpoint of PPPM, frailty, a modifiable risk factor, could be associated with early prediction and primary prevention of mortality, further providing personalized intervention through primary care, which prompts a paradigm shift from real-time diagnostics and treatment to prediction and prevention.

Strengths and limitations

Our study has several major strengths. First, to our knowledge, this study is the first analysis to explore the mediating effect of frailty on the association between LTL and mortality and provides suggestive evidence. Second, the counterfactual framework of mediation analysis was performed in a large sample population to estimate the role of frailty in the association between LTL and mortality. The results might help health policymaker plan more targeted interventions. Furthermore, multiple sensitivity analyses were applied to check the robustness of our results.

Nevertheless, this study also has several limitations. First, as some items for defining FI or FP and covariates were self-reported, the mediators and confounders might have been misclassified. Second, we could not take into account changes in frailty during the follow-up period because some items to evaluate frailty were assessed at baseline. Third, we could only obtain the measurement of LTL at baseline, so it was impossible to assess how its changes affect mortality and the mediated proportion of frailty. Finally, this study was based on an observational design, and no in vivo or in vitro experiments were undertaken. Although we adjusted for multiple covariates, residual confounding may still exist due to the nature of the observational study, and causation cannot be detected. The current findings only offer indicative evidence based on statistical analysis, providing insights and directions for future research. These findings should be verified in the future using randomized controlled trials and genomic analysis.

Conclusions and expert recommendations

To summarize, in this large prospective study, it was verified that longer LTL was significantly associated with reduced risk of all-cause mortality and CVD-specific mortality and that frailty was related to elevated risk of all-cause, CVD-specific, and cancer-specific mortality. In addition, it showed that frailty mediated approximately 10% of the associations. The findings suggest LTL and frailty measuring might be useful to predict the risk of mortality, provide early screening of the high-risk population, and thus impose personalized interventions for modifiable frailty items, which is important to reduce mortality and extend life expectancy.

Predictive medical approach

Based on time-dependent COX analyses and mediation analyses, this study comprehensively investigated the effect of LTL and frailty on mortality and identified that LTL and frailty may be excellent predictors for the risk of mortality, which could provide early screening of high-risk population.

Targeted prevention

Apart from the early predictive medical approach, this study revealed the potential possibility of frailty on preventing the risk of mortality in primary targeted prevention, which might help health policymakers plan more targeted interventions. Frailty has the advantages of being modifiable and cost-effective.

Personalized treatments

Furthermore, targeting the high-risk population identified by LTL or frailty, it is recommended for primary healthcare provider to screen frail individuals timely and impose personalized interventions to modify frailty status. For example, individuals with chronic diseases should pay more attention to self-management and individuals with poor life styles should improve their lifestyles. Additionally, screening programs are suggested to fucus on the elder who were susceptible to be frail for maximum economic health benefits. Targeted primary prevention or personalized interventions were of great significance to control and realize the reduction of mortality.

Overall, our results suggest that frailty mediates the relationships of LTL with all-cause and CVD-specific mortality, which might provide novel targets for PPPM of mortality. Nevertheless, since this research was simply an exploratory mediation analysis and future studies should investigate it further to confirm our results.

PPPM innovation highlights

A. Working hypothesis in the framework of PPPM

We searched PubMed from inception to November 3, 2023, using the following search terms: title/abstract—(telomere* OR frailty) AND (telomere* OR mortality*) AND (frailty OR mortality), with no date or language restrictions. We hypothesized that frailty mediates the association between leucocyte telomere length and mortality. LTL and frailty could serve as biomarkers for predicting mortality and new targets for reducing mortality. Effective identification of frail individuals can provide early screening of the high-risk population, personalized intervention, and timely prevention of mortality. From the perspective of PPPM/3PM, if frailty mediates the association between LTL and mortality, there would be a novel target for PPPM to reduce mortality.

B. Innovation towards the following

1. Predictive approach

We performed time-dependent COX analyses to investigate the effect of LTL and frailty on mortality (all-cause, CVD-specific, and cancer-specific mortality). Based on the mediation analyses, we explored the role of frailty in the association between telomere length and mortality, a potential target to prevent and reduce mortality.

• 2.Targeted prevention

It’s confirmed that LTL and frailty can be measured and calculated to predict the risk of mortality. In addition, frailty has the advantage of being modifiable. Therefore, personalized interventions are supposed to be tailored based on each individual’s unique profile of frailty items. For example, individuals with chronic diseases should pay more attention to self-management and individuals with poor life styles should improve their lifestyles to have a bigger profit.

• 3.Personalization of medical services

Accumulating evidence indicated that longer LTL or frailty could effectively predict the risk of mortality. This study also showed longer LTL or frailty significantly decreased the risk of all-cause and CVD-specific mortality and frailty mediated approximately 10% of the association between them, indicating their essential role in predictive, preventive, and personalized medicine. It is recommended for primary healthcare provider to screen frail individuals timely and impose personalized interventions to modify frailty status. Additionally, screening programs are suggested to fucus on the elder who were susceptible to be frail for maximum economic health benefits.

C. How does the presented innovation go beyond the state of the art contributing to the paradigm shift from reactive medicine to PPPM?

In conclusion, LTL and frailty have the potential value on prediction, prevention, and personalization medicine of mortality. Maintaining longer LTL or non-frail status would increase the quality of life and extend life expectancy and thus reduce mortality. Additionally, it is possible to change frailty status and reduce mortality risk through personalized interventions for frailty items in primary prevention or primary healthcare, implying novel targets for mortality prevention and reduction.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

CI

Confidence interval

CVD

Cardiovascular diseases

FI

Frailty index

FP

Frailty phenotype

HR

Hazard ratio

IQR

Interquartile range

LTL

Leucocyte telomere length

NDE

Natural direct effect

NIE

Natural indirect effect

PPPM/3PM

Predictive, preventive, and personalized medicine

SD

Standard deviation

TDI

Townsend deprivation index

TE

Total effect

Author contribution

LW and YW contributed to the study conception and design. Analysis and interpretation of data was performed by XJ, WS, JZ, and HL. Drafting of the manuscript was performed by XJ, WS, QZ, and XM. YW did critical revision of the manuscript for important intellectual content. All authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by the National Key R&D Program of China-European Commission Horizon 2020 (2017YFE0118800-779238) and Beijing Talents Project (2020A17).

Availability of data and material

This research was conducted using the UK Biobank study under Application Number 95259. Data from UK Biobank are available on application at www.ukbiobank.ac.uk/register-apply.

Code availability

Data are available on reasonable request from the corresponding author.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval

Ethical approval of the UK Biobank was obtained from the National Health Service National Research Ethics Service, and all participants provided written informed consent.

Consent to participate

Written informed consent was obtained from all participants.

Consent for publication

All authors gave their consent for publication.