The data used in this study were obtained from the UK Biobank (ID: 107175). The UK Biobank is a large-scale prospective cohort study that recruited over 500,000 volunteers across the United Kingdom between 2006 and 2010 [13]. This study received ethical approval from the North West Multi-Centre Research Ethics Committee (Ref: 11/NW/0382), and all participants provided informed consent by signing a consent form. This study was reported in accordance with the STROBE guidelines.

Of the 502,163 participants in the UK Biobank, we excluded 460 participants with baseline AAA, 177,504 participants missing biomarkers required for the KDM-BA and PhenoAge algorithms, 11,531 participants missing TL measurements, and 1022 participants with missing covariate data. A total of 311,646 participants were included in the final analysis. The participant selection process is shown in Supplementary Fig. 1.

The outcome variable of this study was the occurrence of AAA, which was determined based on hospital admission and death registration records (Supplementary Table 1). The diagnosis of AAA was identified using the Tenth Revision of the International Classification of Diseases (ICD-10 codes: I71.3, I71.4) and the Office of Population, Censuses and Surveys: Classification of Interventions and Procedures (OPCS-4 codes: L18*, L19*, L254, L27*, L28*, L464), as used in prior studies [14]. The study period began when participants entered the cohort and ended upon the diagnosis of AAA, loss to follow-up, death, or completion of follow-up, whichever occurred first. The follow-up cut-off date was September 30, 2021, for England and Wales, and October 31, 2021, for Scotland.

TL was measured using a multiplex quantitative polymerase chain reaction method, with samples obtained from peripheral blood leukocytes. Specifically, TL was quantified relative to the ratio of telomere repeat copy number to single-copy gene (T/S), and the data were log-transformed and Z-standardized. Detailed information on the measurement and adjustment of TL can be found in previous studies [15].

The KDM-BA and PhenoAge algorithms [16, 17] are widely recognized biological age measures based on clinical parameters. These algorithms were developed using optimal training models derived from the National Health and Nutrition Examination Survey (NHANES) and have been further validated with data from the UK Biobank. Using the R package “BioAge” (https://github.com/dayoonkwon/BioAge), KDM-BA was calculated based on forced expiratory volume in one second, systolic blood pressure, and seven blood chemistry parameters: albumin, alkaline phosphatase, blood urea nitrogen, creatinine, C-reactive protein (CRP), glycated hemoglobin, and total cholesterol. PhenoAge was calculated using nine hematological indicators: albumin, alkaline phosphatase, creatinine, CRP, glucose, mean corpuscular volume, red cell distribution width, white blood cell count, and lymphocyte percentage. We then regressed the calculated biological age values against the chronological age at the time of biomarker measurement and computed the residuals [18]. These residuals were termed “age acceleration” (AA) values, which were used to assess biological ageing. The AA values were then standardized with a mean of 0 and a standard deviation of 1 for comparison in subsequent analyses. Further details can be found in Supplementary Method 1.

The covariates considered in this study included age, sex, ethnicity, education level, Townsend Deprivation Index (TDI), CRP, body mass index (BMI), smoking and alcohol consumption status, physical activity, sleep and dietary patterns, as well as a medical history of hyperlipidemia, hypertension, and diabetes. Detailed definitions and descriptions of these covariates can be found in Supplementary Method 2 and Supplementary Tables 2,3.

Based on the Kolmogorov-Smirnov test (all p-values 0.001) and quantile-quantile (Q-Q) plots, the continuous variables—including age, TDI, BMI, and CRP—were found not to follow a normal distribution and were therefore expressed as medians [interquartile range (IQR)]. Group comparisons were performed using the Mann-Whitney U test or Kruskal-Wallis H test. Categorical variables were represented as frequencies (percentages), and group comparisons were conducted using chi-square tests.

TL, KDM-BA, and PhenoAge acceleration were treated as continuous variables or categorized based on quartiles, and their associations with AAA occurrence were assessed using Cox proportional hazards models, with results presented as hazard ratios (HR) and 95% confidence intervals (95% CI). The Schoenfeld residual test confirmed no violations of the proportional hazards assumption. Model 1 adjusted for age, sex, ethnicity, education level, and TDI. Model 2 further adjusted for smoking and drinking status, physical activity, sleep and dietary patterns, BMI, CRP, and medical history of hypertension, hyperlipidemia, and diabetes. Additionally, based on Cox regression Model 2, a restricted cubic spline (RCS) with three knots was used to analyze the dose-response relationship between biological ageing and AAA risk.

We included only participants of White ancestry and constructed a polygenic risk score (PRS) to assess individuals’ genetic susceptibility to AAA [19] (Supplementary Method 3). We then evaluated the joint effect of genetic susceptibility and biological ageing on the risk of AAA. Participants were categorized into 12 groups based on their levels of biological ageing and genetic susceptibility. Using the group with the lowest levels of both biological ageing and genetic susceptibility as a reference, we estimated the HR and 95% CI for AAA occurrence in the other groups, after adjusting for covariates in Model 2. Stratified analyses were conducted to assess the separate effect of biological ageing on the incidence of AAA at different levels of genetic susceptibility. The likelihood ratio test was used to evaluate the interaction between biological ageing and genetic susceptibility.

Subgroup analyses were also conducted to assess potential modifying effects on the association between biological ageing and the risk of AAA. Sensitivity analyses included: (1) Excluding participants who developed AAA within the first year of follow-up; (2) Using the Fine-Gray model to account for mortality as a competing risk. All analyses were performed in R (version 4.3.2, R Foundation for Statistical Computing, Vienna, Austria), with p-values 0.05 considered statistically significant.

This study included a total of 311,646 participants (Supplementary Fig. 1), of whom 168,249 (53.99%) were women. The majority of participants were White, comprising 294,574 individuals or 94.52% of the total cohort (Supplementary Table 4). The median age was 58 years (IQR = 50–63 years), with median values of KDM-BA and PhenoAge recorded at 49.82 years (IQR = 40.90–58.26) and 45.83 years (IQR = 38.10–52.54), respectively. KDM-BA (r = 0.67) and PhenoAge (r = 0.86) were strongly positively correlated with chronological age, whereas TL showed a weak negative correlation (r = –0.20, Supplementary Fig. 2).

Baseline characteristics, stratified by quartiles of TL, PhenoAge acceleration, and KDM-BA acceleration, are summarized in Table 1 and Supplementary Tables 5,6. Individuals with longer TL had a higher proportion of women, a higher level of education, higher TDI, lower BMI and CRP, and a higher proportion of those maintaining a healthy lifestyle (never smoking, never drinking, regular physical activity, a healthy sleep pattern, and a healthy diet). They also had a lower proportion of comorbid hypertension, diabetes, and hyperlipidemia. Similarly, participants with higher KDM-BA acceleration and higher PhenoAge acceleration had higher TDI, lower educational attainment, higher CRP levels, and a higher prevalence of comorbidities such as hypertension, diabetes, and hyperlipidemia.

During a median follow-up of 12.54 years (IQR = 11.87–13.16), a total of 1339 (4.29‰) incident cases of AAA were reported. Overall, accelerated biological ageing was significantly positively associated with an increased long-term risk of AAA (Table 2). For each standard deviation increase in TL, the risk of AAA decreased by 17% (HR = 0.83, 95% CI = 0.79–0.88); for each standard deviation increase in KDM-BA acceleration, the risk increased by 21% (HR = 1.21, 95% CI = 1.12–1.29); and for each standard deviation increase in PhenoAge acceleration, the risk increased by 40% (HR = 1.40, 95% CI = 1.32–1.48). Participants with longer TL had a significantly lower risk of AAA compared with those with shorter TL, while those with higher KDM-BA acceleration or higher PhenoAge acceleration had a significantly higher risk of AAA compared with those with lower KDM-BA acceleration or lower PhenoAge acceleration. Fig. 1 demonstrates the potential dose-response relationships between biological ageing and the risk of AAA. The results indicated a linear association between TL, KDM-BA acceleration, PhenoAge acceleration, and the occurrence of AAA (p-value for non-linearity $>$0.05).

Fig. 1.

Association between biological ageing and the risk of AAA using restricted cubic splines model with three knots. Panels A-C depict the dose-response relationships between telomere length (A), KDM-BA acceleration (B), and PhenoAge acceleration (C) and the risk of AAA, respectively. Abbreviation: AAA, abdominal aortic aneurysm.

The results in Supplementary Table 7 indicated that the constructed PRS score effectively predicted the risk of the occurrence of AAA. After further adjusting for PRS in Models 1 and 2, the significant association between biological ageing indicators and AAA risk persisted (Supplementary Table 8).

The joint effects of biological ageing and genetic susceptibility on the risk of AAA are shown in Fig. 2A–C. The risk of AAA increased with accelerated biological ageing and higher levels of genetic susceptibility (p-value for trend 0.001). Compared to participants with the lowest PRS and biological ageing, those with the highest PRS and biological ageing exhibited the highest risk of AAA (high PRS and Q1 of TL: HR = 1.81, 95% CI = 1.31–2.49; high PRS and Q4 of KDM-BA acceleration: HR = 2.43, 95% CI = 1.79–3.28; high PRS and Q4 of PhenoAge acceleration: HR = 3.27, 95% CI = 2.21–4.85). At different levels of genetic susceptibility, higher levels of biological ageing were positively associated with the occurrence of AAA (Fig. 2D–F). No significant interaction was observed between genetic susceptibility and biological ageing indicators regarding the risk of AAA (p-value for interaction $>$0.05).

Fig. 2.

Separate and joint association of biological ageing with the risk of AAA in individuals with different levels of genetic susceptibility. Panels A–C show the joint effects of telomere length (A), KDM-BA acceleration (B), and PhenoAge acceleration (C) with genetic susceptibility, respectively; Panels D–F show the separate effects of telomere length (D), KDM-BA acceleration (E), and PhenoAge acceleration (F) across different levels of genetic susceptibility, respectively. The result was presented as HR (95% CI) using the Cox proportional hazards model adjusted for age, sex, ethnicity, Townsend deprivation index, education levels, body mass index, C-reactive protein, smoking and drinking status, physical activity, sleep and diet patterns, history of hyperlipemia, hypertension, and diabetes mellitus.

As shown in Table 3, subgroup analyses indicated that the effect of longer TL in reducing the incidence of AAA was more significant in men than in women (p-value for the interaction between sex and TL = 0.018). For other stratification factors, no significant differences were observed in the associations between biological ageing indicators and AAA occurrence (p-value for the interaction $>$0.05, Supplementary Tables 9,10). Sensitivity analyses, including the exclusion of participants who reported AAA in the first year (Supplementary Table 11) and the application of Fine-Gray competing risk regression models (Supplementary Fig. 3), consistently supported the strong association between high levels of biological ageing and an increased incidence of AAA.