Blood biomarker profiles and exceptional longevity: comparison of centenarians and non-centenarians in a 35-year follow-up of the Swedish AMORIS cohort

Comparing biomarker profiles measured at similar ages, but earlier in life, among exceptionally long-lived individuals and their shorter-lived peers can improve our understanding of aging processes. This study aimed to (i) describe and compare biomarker profiles at similar ages between 64 and 99 among individuals eventually becoming centenarians and their shorter-lived peers, (ii) investigate the association between specific biomarker values and the chance of reaching age 100, and (iii) examine to what extent centenarians have homogenous biomarker profiles earlier in life. Participants in the population-based AMORIS cohort with information on blood-based biomarkers measured during 1985–1996 were followed in Swedish register data for up to 35 years. We examined biomarkers of metabolism, inflammation, liver, renal, anemia, and nutritional status using descriptive statistics, logistic regression, and cluster analysis. In total, 1224 participants (84.6% females) lived to their 100th birthday. Higher levels of total cholesterol and iron and lower levels of glucose, creatinine, uric acid, aspartate aminotransferase, gamma-glutamyl transferase, alkaline phosphatase, lactate dehydrogenase, and total iron-binding capacity were associated with reaching 100 years. Centenarians overall displayed rather homogenous biomarker profiles. Already from age 65 and onwards, centenarians displayed more favorable biomarker values in commonly available biomarkers than individuals dying before age 100. The differences in biomarker values between centenarians and non-centenarians more than one decade prior death suggest that genetic and/or possibly modifiable lifestyle factors reflected in these biomarker levels may play an important role for exceptional longevity. Supplementary Information The online version contains supplementary material available at 10.1007/s11357-023-00936-w.


Additional methods of statistical analysis
In multiple imputation process, we created 100 imputed data sheets assuming missing at random.To avoid model misspecification, random forests were used to predict missing values [9].The predictors in the random forest were all available biomarker values, age, sex, and specific comorbidities.The imputation was conducted using the mice and randomForest packages in R [10,11].Rubin's rule was used to combine the results of the 100 individual imputations.In the first step investigating differences in the biomarkers' destributions, quantile regression were built to compare 75th quantile of biomarkers with adjustment age and sex.We compared the distribution of biomarker values using the 10th quantile, 25th quantile, median (50th quantile), 75th quantile, and 90th quantile among three groups: people who died before their 90th birthday (sexagenarians, septuagenarians, and octogenarians), people who died between their 90th birthday and 100th birthday (nonagenarians), and people reaching their 100th birthday (centenarians).Those quantiles were also compared with the normal range of each biomarker based on previously established evidence (Supplemental table 1).In a subsample with a repeated creatinine measurement within 5 years of the first measurement, we further compared quantiles and means (standard deviation) of first and second measuremed creatinine as well as the mean (standard deviation) change in supplemental figure 4. As a sensitivity analysis, we also analyzed CRP (an inflammatory marker) which was measured for a subset of participants, the iron/TIBC ratio as an indicator of iron deficiency, and the ASAT/ALAT ratio as a marker of liver status in the supplemental figure 7 and table 6 [12,13].We compared mean values and 95% confidence intervals (CI) between centenarians and non-centenarians using ordinary least squares with estimated marginal means [14].In the third step investigating the variation in biomarker profiles, we used K-medians clustering since we observed some outliers in the biomarker distributions.This study followed the steps of clustering analyses using multiply imputed data sets presented previously [15].We used forward sequential selection to reduce the variables used in the clustering process [16].The optimal number of clusters was explored and chosen based on the most frequently selected number of clusters in 100 imputed data sets using CritCF [15].We included all biomarkers that appeared at least once in any K-medians variable selection of 100 multiply imputed data sets due to sufficient sample size in our data [15].Some participants can be allocated to different clusters when different imputed data sets are used.The algorithm to determine cluster membership in our final model was based on all scenarios estimated with 100 imputed data sets.After the clustering procedure, baseline characteristics were compared between clusters and survival differences were also examined between clusters using Kaplan-Meier curves and Cox proportional hazards model.10: Quantiles (10th, 25th, 50th, 75th, 90th) of selected biomarkers among each centenarian cluster and non-centenarians restricting to participants whose blood was sampled before age 80 years.Green areas show each biomarker's normal range.Multiply imputed data were used and 667 centenarians and 30,796 non-centenarians were included.TC, total cholesterol; ALAT, alanine aminotransferase; ASAT, aspartate aminotransferase; GGT, gamma-glutamyl transferase; ALP, alkaline phosphatase; TIBC, total iron-binding capacity.

Supplemental figure 3 :Supplemental figure 4 :
Biomarkers' means and 95% confidence intervals standardized using mean and standard deviation observed in the total study population for centenarians and non-centenarians.Multiply imputed data were used and 44,636 participants were included.TC, total cholesterol; ALAT, alanine aminotransferase; ASAT, aspartate aminotransferase; GGT, gamma-glutamyl transferase; ALP, alkaline phosphatase; TIBC, total iron-binding capacity; M, mean in total study population; SD, standard deviation in total study population Quantiles (10th, 25th, 50th, 75th, 90th) and means (standard deviation) of first and second creatinine measurements, and their changes between centenarians and non-centenarians.SD, standard deviation.Creatinine was measured twice in 51.6% (N=21,433) of participants, which were included in the analysis.

Supplemental figure 5 :
Association between biomarker quintiles and the chance of becoming a centenarian: logistic regression using complete case data.
areas show each biomarker's normal range based on commonly-used clinical thresholds (see Supplemental table 1 for further details).Complete case data were used and 26,666 participants were include.TC, total cholesterol; ALAT, alanine aminotransferase; ASAT, aspartate aminotransferase; GGT, gamma-glutamyl transferase; ALP, alkaline phosphatase; TIBC, total iron-binding capacity.

Supplemental table 6: Association of CRP, ASAT/ALAT, and iron/TIBC quintile with the chance of becoming a centenarian: logistic regression.
-reactive protein; ALAT, alanine aminotransferase; ASAT, aspartate aminotransferase; TIBC, total iron-binding capacity; OR, odds ratio; CI, confidence interval Each model was adjusted by age, sex, and CCI.*ASAT/ALAT and iron/TIBC were analyzed using multiply imputed data.0.5% of participants excluded due to difference in assessment of ASAT and ALAT or iron and TIBC.†Complete case data were used when analyzing CRP.CRP was measured in 46% (N=20,455) of participants, which were included.

Supplemental figure 9: Comparison of suvival probability between clusters: Kaplan-Meier survival curves and Cox proprotional hazard model.
HR were estimated using a crude Cox model.Multiply imputed data were used and 1,224 centenarians were included.HR, hazard ratio; CI, confidence interval