Supplementary material and methods
Additional biomarkers not included in previous publications were Leptin, Adiponectin,
Transforming growth factor beta (TGF-beta) and Insulin-like growth factor-binding
proteins 1 and 3 (IGFBP1 and IGFBP3). For all of these biomarkers, their corresponding
Sandwich ELISA R&D DuoSet Development System protocols (R&D Systems, Abingdon,
UK) were tested and suitably modified to be transferred onto a 384-well plate platform
with the use of an EpMotion 5075 Liquid Handling robot. All colorimetric measurements
were performed at 450nm with wavelength correction at 550nm on a FLUOstar Omega
microplate reader (BMG Labtech, Aylesbury, UK) applying a four-parameter logistic (4PL) curve fit to the respective standard curves. Table S1 describes specific details of the
protocols.
For details of previously reported biomarkers not included in 24, see the following
references: S1 for mitochondrial haplogroups, S2 for DNA methylation, and S3 for APOE
genotype.
1
Supplementary references
S1. Collerton J, Ashok D, Martin-Ruiz C, et al. Frailty and mortality are not influenced
by mitochondrial DNA haplotypes in the very old. Neurobiol Aging. 2013; 34:
2889 e1-4.
S2. Collerton J, Gautrey H, Otterdijk S, et al. Acquisition of aberrant DNA
methylation is associated with frailty in the very old: findings from the
Newcastle 85+ Study. Biogerontology. 2014; 15: 317-28.
S3. Deelen, J; Beekman, M; Uh, H-W; Broer, L, et al., Genome-wide association metaanalysis of human longevity identifies a novel locus conferring survival beyond
90 years of age. Hum. Mol. Genet. 2014; 23: 4420-32.
2
Supplementary Figures and Table legends
Figure S1. Kaplan-Meier survival curves for dichotomized biomarkers. Kaplan-Meier
survival curves of the biomarkers split by the two groups using the cut points (Table 1).
Red lines show survival curves for the individuals who are “at higher risk” and the blue
lines are for “at lower risk” individuals.
Figure S2. Kaplan-Meier survival curves of the FI-B calculated from biomarkers (the
number is indicated on the top of each panel) randomly chosen from the total 40 and
stratifying them in four groups with the same cut points as in Figure 3. The sampling is
repeated 100 times.
Table S1. Specific details of the protocols for additional biomarkers.
Table S2. Key socio-demographic and health parameters reported for samples with FICD and FI-B. Comparison of the FI-CD sample with FI-B sample.
Table S3. The area under the receiving operating characteristic (AUC) with 95%
confidence intervals for different versions of the FI and Fried phenotype in the
subsample of participants for whom the Fried phenotype was available (n=552).
3
Figure S1. Kaplan-Meier survival curves of the individual biomarkers split by two groups
using the cut points chosen to achieve the best separation of survival curves between
people with and without the deficit and minimizing the p-value of the log rank test
(Table 1). Red lines show survival curves for the group “at higher risk” and the blue lines
are for “at lower risk” individuals.
INFLAMMATION
Survival probability
Cytomegalovirus serology, p=0·327
High sensitivity C-reactive protein, p<0·001
1
1
0.5
0.5
0.5
0
0
500
1000
1500
2000
2500
0
0
500
days-to-death
1000
1500
2000
2500
0
IL-6, post stimulation, p=0·92
TNF-alpha, basal, p=0·646
0.5
0.5
0.5
1000
1500
2000
2500
0
0
500
Leptin, p=0·023
1
1500
2000
2500
500
1000
1500
2000
2500
days-to-death
0
0
500
0
2000
2500
1000
1500
2000
2500
Homocysteine, p=0·006
1
0.5
0
1500
days-to-death
Adiponectin, p=0·012
1
0.5
0
1000
days-to-death
days-to-death
1000
TNF-alpha, post stimulation, p=0·116
1
500
500
days-to-death
1
0
0
days-to-death
1
0
Survival probability
IL-6, basal, p=0·399
1
0.5
0
500
1000
1500
days-to-death
2000
2500
0
0
500
1000
1500
2000
2500
days-to-death
Albumin, p<0·001
1
0.5
0
0
500
1000
1500
2000
2500
days-to-death
4
Figure S1 continued
HAEMATOLOGICAL
Haemoglobin, p<0·001
1
Survival probability
0.5
0
Platelets, p=0·003
1
0.5
0
500
1000
1500
2000
2500
0
0.5
0
500
Neutrophils, p=0·007
1
2000
2500
0
500
1000
1500
2000
2500
500
0
1000
1500
2000
0.5
0
500
1000
1500
2000
2500
0
0
500
1000
1500
2000
days-to-death
Eosinophils, p=0·15
1
0.5
0.5
0
500
1000
1500
days-to-death
2000
2500
0
2500
Monocytes, p=0·14
days-to-death
Basophils, p=0·156
1
0
days-to-death
0.5
0
0
1
days-to-death
Survival probability
1500
Lymphocytes, p=0·028
1
0.5
0
1000
days-to-death
days-to-death
0
White blood cells, p=0·035
1
0
500
1000
1500
2000
2500
days-to-death
5
2500
Figure S1 continued
IMMUNOSENESCENCE
CD4 T cells, p=0·029
1
0.5
0.5
Survival probability
1
0
0
500
1000
1500
2000
2500
0
1
0.5
0
500
days-to-death
0.5
0
0
500
1000
1500
2000
2500
Survival probability
2000
2500
Memory/Naive CD4 T cell ratio, p=0·034
500
1000
1500
days-to-death
0
0.5
0.5
0
500
1000
1500
2000
2500
2000
2500
0
1500
2000
2500
0
0
500
1000
1500
2000
2500
days-to-death
Memory/Naive CD8 T cell ratio, p=0·352
1
1000
CD4/CD8 T cells ratio, p=0·005
1
0
500
days-to-death
Memory CD4 T cells, p=0·057
0.5
0
0
days-to-death
0.5
0
1500
1
days-to-death
1
1000
days-to-death
Senescent memory CD4 T cells, p=0·079
1
CD8 TEMRA T cells, p=0·014
CD8 T cells, p=0·015
Memory/Naïve B cell ratio, p=0·026
1
0.5
0
500
1000
1500
days-to-death
2000
2500
0
0
500
1000
1500
2000
2500
days-to-death
6
Figure S1 continued
CELLULAR AGEING/ OXIDATIVE STRESS
Survival probability
Telomere length, p=0·118
DNA repair, p=0·078
1
1
0.5
0.5
0
0
500
1000
1500
2000
2500
0
0.5
0
500
days-to-death
2000
2500
0
500
1000
1500
00
2000
2500
0
0
500
1000
1500
2000
2500
0
iPF2alpha-VI (LC/MS/MS), p=0·10
0.5
0.5
1000
1500
500
2000
2500
0
0
500
1000
1500
days-to-death
1000
1500
2000
2500
iPF2alpha-III (AutoDELFIA), p=0·258
0.5
days-to-death
2500
days-to-death
1
500
2000
0
1
0
1500
0.5
days-to-death
iPF2alpha-III (LC/MS/MS), p=0·053
1000
IGFBP3, p=0·357
1
1
0
500
days-to-death
0.5
days-to-death
Survival probability
1500
IGFBP1, p=0·001
1
0.5
0
1000
days-to-death
TGF beta, p=0·003
1
DNA damage/DNA repair ratio, p=0·063
1
2000
2500
0
0
500
1000
1500
2000
days-to-death
7
2500
Figure S1 continued
GENETIC/EPIGENETIC
Mitochondrial DNA haplogroup, p=0·031
1
Survival probability
0.5
0
APOE genotype (4), p=0·024
1
0.5
0
500
1000
1500
2000
2500
0
0.5
0
500
1000
1500
days-to-death
days-to-death
CPG island DNA methylation, p=0·67
1
2000
2500
0
0
500
1000
1500
2000
days-to-death
line1 DNA methylation, surrogate for genomewide DNA methylation, p=0·63
1
0.5
0
0
500
1000
1500
2000
2500
days-to-death
8
2500
Figure S2. Kaplan-Meier survival curves of the FI-B calculated from biomarkers (the
number is indicated on the top of each panel) randomly chosen from the total 40 and
stratifying them in four groups with the same cut points as in Figure 2. The sampling is
repeated 100 times.
9
Table S1. Specific details of the protocols for additional biomarkers.
Biomarker
Leptin
Adiponectin
TGF-beta
IGFBP1
IGFBP3
R&D Duoset
DY398
DY1065
DY240
DY871
DY675
Sample
Serum
Serum
Serum
Serum
Serum
Dilution
1:40
1:6000
1:20
1:100
1:200
Activation with
1N HCl
Reagent Diluent
Neutralisation
Dilution
Conditions
Reagent Diluent
Reagent Diluent
DY995 including
DY995 including
1%BSA
1%BSA
with 1.2N
NaOH/0.5M
HEPES
Reagent Diluent
DY004 including
DY004 including
5% Tween 20
5% Tween 20
and 2% Normal
Goat Serum
Reagent Diluent
DY997 including
0.05% Tween 20
Capture
Antibody
1.33ug/ml
1.0ug/ml
2.0 µg/ml
4.0ug/ml
4.0ug/ml
4.16ng/ml
1.0ug/ml
300 ng/ml
400ng/ml
200ng/ml
Standard
0.5ng/ml to
0.6ng/ml to
2ng/ml to
2ng/ml to
8ng/ml to
Curve Range
156.25pg/ml
187.5pg/ml
62.5 pg/ml
62.5pg/ml
250pg/ml
Inter-assay CV
4.21%
7.48%
3.32%
4.89%
4.43%
Intra-assay CV
2.72%
6.06%
6.03%
5.42%
4.52%
concentration
Detection
Antibody
concentration
Table S2. Key socio-demographic and health parameters –reported for samples with FICD and FI-B. Comparison of the FI-CD sample with FI-B sample.a
10
Table S2. Key socio-demographic and health parameters reported for samples with FICD and FI-B. Comparison of the FI-CD sample with FI-B sample.
Female
Years of Education
12+
10-11
0-9
Smoking Status
Never smoker
Ex-smoker
Current smoker
Body mass index (kg/m²)
<18.50
18.50-24.99
25.00-29.99
≥30.00
Cognitive function
Normal (26-30)
Mildly impaired (22-25)
Moderately impaired (18-21)
Severely impaired (0-17)
FI-CD (mean, SD)
a The
b
p valueb
Sample with FI-CD
n=811
61·6 (500)
Sample with FI-B
n=777
60·9 (473)
12·4 (99)
23·4 (187)
64·2 (512)
12·6 (97)
23·5 (180)
63·9 (490)
0·966
0·982
0·921
35·4 (286)
59·0 (477)
5·7 (46)
35·1 (272)
59·1 (459)
5·8 (45)
0·941
0·975
0·984
6·6 (47)
51·4 (368)
32·5 (233)
9·5 (68)
6·6 (46)
51·1 (356)
33·1 (231)
9·2 (64)
0·999
0·936
0·891
0·953
74·0 (582)
15·0 (118)
4·9 (39)
6·1 (48)
0·22 (0.12)
74·3 (564)
15·3 (116)
5·0 (38)
5·4 (41)
0·21 (0.12)
0·908
0·949
0·984
0·888
0·605
0·822
data are % (n) except where specified; denominators vary due to missing values.
p values are calculated from the test of proportions except for the last row where
Student t-test was used for no significant differences between FI-CDs in the two
samples.
11
Table S3. The area under the receiving operating characteristic (AUC) with 95%
confidence intervals for different versions of the FI and Fried phenotype in the
subsample of participants for whom the Fried phenotype was available (n=552).
FI-B
FI-CD
FI-CD + FI-B
FI-CD + FI-B + sex
FI-CD + FI-B + sex + Fried
phenotype
Fried phenotype
AUC
0.61
0.64
0.68
0.72
0.72
95% CI
0.57 – 0.66
0.60 – 0.69
0.63 – 0.73
0.68 – 0.76
0.68 – 0.77
0.58
0.53 – 0.63
12