Conceived and designed the experiments: PGS LMM A. MacIntyre PJ GDB HB JC KAD IF A. McConnachie A. McGinty JSM KM NS CT YNV CJP. Performed the experiments: LMM A. MacIntyre PGS. Analyzed the data: LMM PGS PJ A. McConnachie. Contributed reagents/materials/analysis tools: PGS LMM A. MacIntyre PJ GDB HB JC KAD IF A. McConnachie A. McGinty JSM KM NS CT YNV CJP. Wrote the paper: PGS LMM A. MacIntyre PJ GDB HB JC KAD IF A. McConnachie A. McGinty JSM KM NS CT YNV CJP.
The authors have declared that no competing interests exist.
It has previously been hypothesized that lower socio-economic status can accelerate biological ageing, and predispose to early onset of disease. This study investigated the association of socio-economic and lifestyle factors, as well as traditional and novel risk factors, with biological-ageing, as measured by telomere length, in a Glasgow based cohort that included individuals with extreme socio-economic differences.
A total of 382 blood samples from the pSoBid study were available for telomere analysis. For each participant, data was available for socio-economic status factors, biochemical parameters and dietary intake. Statistical analyses were undertaken to investigate the association between telomere lengths and these aforementioned parameters.
The rate of age-related telomere attrition was significantly associated with low relative income, housing tenure and poor diet. Notably, telomere length was positively associated with LDL and total cholesterol levels, but inversely correlated to circulating IL-6.
These data suggest lower socio-economic status and poor diet are relevant to accelerated biological ageing. They also suggest potential associations between elevated circulating IL-6, a measure known to predict cardiovascular disease and diabetes with biological ageing. These observations require further study to tease out potential mechanistic links.
Gompertz (1825) first described ageing as an increase in the likelihood of mortality with increasing chronological age
Variation in the rate of biological ageing reflects the cumulative burden of genetic, metabolic and environmental stressors, resulting in oxidative damage and elevated inflammatory processes
Suitable and validated biomarkers for analysing biological ageing in this context are limited in number. Despite cell cycle inhibitor transcript levels providing an accurate indication of organ and T cell biological age
Telomeres are nucleoprotein complexes at chromosome ends, consisting of (TTAGGG)n direct repeats bound to a range of proteins involved in maintaining cellular stability and viability (reviewed in
Although telomere length is inversely related to chronological age in humans, there is considerable inter-individual variation in telomere length at any specific age
Recent human data have now established telomere attrition as a major risk factor for numerous diseases, including cardiovascular disease (CVD), hypertension, diabetes and end stage renal disease
One hypothesis for the increased disease prevalence in these communities is underlying chronic inflammation (elevated CRP or IL-6), a known component and predictor of CVD and diabetes
A link between accelerated biological ageing and socio-economic status has previously been reported by some
We have chosen to evaluate the contribution of socio-economic factors to biological age, as measured by telomere length, in the extreme setting of the pSoBid cohort, to determine to what extent this in turn affects risk factors for ill health.
Telomere lengths were determined in PBLs by Q-PCR for 382 individuals. The SES and lifestyle factors investigated are shown in
The relationship between age, gender and biological ageing was investigated by estimating the percentage change in telomere length associated with a decade increase in age and male gender. Age was strongly negatively associated with telomere length, each decade predicting a 4.8% decrease in telomere length (p = 0.002). No significant difference in telomere length was observed between males and females. There was also no difference between affluent and deprived groups either in telomere length or age-related telomere attrition. All subsequent analyses were adjusted for age, gender and deprivation.
Of the SES and lifestyle factors investigated, only cigarette smoking was associated with an overall reduction in telomere length (6.6% reduction, p = 0.050). However, household income, housing tenure, and diet score were associated with steeper age-related decline in telomere length (
Faster rates of age-related telomere attrition were observed in individuals with an average income less than £25,000 (7.7% vs 0.6% reduction per decade, p = 0.024, (A)), home tenants (8.7% vs 2.2%, p = 0.038 (B)) and a diet score among the lower 50% of scores (7.7% vs 1.8%, p = 0.05 (C)).
% change in telomere length (95% CI) per decade associated with SES & lifestyle factor | |||
SES & Lifestyle Factor | SES & Lifestyle Factor Subgroup NO | SES & Lifestyle Factor Subgroup YES | Differences between subgroups (p-value) |
|
−3.7 (−7.5, 0.3) p = 0.066 | −6.2 (−11.1, −1.1) p = 0.019 | 0.437 |
|
−0.6 (−5.3, 4.2) p = 0.796 |
|
|
|
−1.3 (−5.9, 3.5) p = 0.577 | −7.2 (−11.0, −3.2) p = 0.001 | 0.062 |
|
−2.2 (−6.1, 1.8) p = 0.274 |
|
|
|
−6.0 (−10.5, −1.3) p = 0.013 | −3.6 (−7.5, 0.6) p = 0.092 | 0.426 |
|
−4.8 (−8.2, −1.3) p = 0.008 | −7.4 (−13.6, −0.8) p = 0.028 | 0.482 |
|
−1.8 (−6.1, 2.6) p = 0.417 |
|
|
|
−4.8 (−8.0, −1.3) p = 0.007 | −5.0 (−11.6, 2.0) p = 0.156 | 0.944 |
|
−4.1 (−7.5, −0.4) p = 0.029 | −7.2 (−12.8, −1.2) p = 0.019 | 0.371 |
|
−2.2 (−6.5, 2.2) p = 0.319 | −7.6 (−11.6, −3.4) p = 0.001 | 0.081 |
Individuals with an income < £25,000, housing tenants and those with a lower diet score were shown to have faster rates of age-related telomere attrition than those earning >£25,000, house owners and individuals with higher diet scores. All analyses were adjusted for age, gender and deprivation group.
We investigated the extent to which the associations between telomere attrition and household income, housing tenure, and diet score were independent by adjusting for these interactions. All three interactions were attenuated to a similar degree and were no longer significant, suggesting that these three interactions are correlated but no single factor is driving these interactions (
Associations between biomarkers investigated and telomere length are reported in
Outcome | Percentage Change (95% CI)p-Value |
Systolic BP (mmHg) | 0.2 (−1.1, 1.5) p = 0.798 |
Diastolic BP (mmHg) | 0.3 (−1.0, 1.7) p = 0.642 |
Cholesterol (mmol/l) | 2.4 (0.3, 4.6) p = 0.027 |
HDL cholesterol (mmol/l) | −1.2 (−3.9, 1.6) p = 0.387 |
LDL cholesterol (mmol/l) |
|
Triglycerides (mmol/l) | 2.7 (−2.6, 8.3) p = 0.326 |
Glucose (mmol/l) | 0.8 (−1.0, 2.7) p = 0.363 |
Insulin (mU/l) | −0.5 (−7.5, 6.9) p = 0.884 |
HOMA - IR | −0.1 (−7.5, 8.0) p = 0.989 |
C reactive protein (mg/l) | 0.6 (−9.9, 12.4) p = 0.916 |
Interleukin 6 (pg/ml) |
|
Intercellular adhesion molecule 1 (ng/ml) | −0.3 (−2.9, 2.3) p = 0.796 |
Fibrinogen (g/l) | 0.2 (−1.9, 2.4) p = 0.829 |
von Willebrand factor (IU/dl) | −1.2 (−4.2, 1.8) p = 0.416 |
D-dimer (ng/ml) | −1.7 (−7.4, 4.4) p = 0.583 |
Individuals with longer telomeres had increased levels of total cholesterol (2.4% increase, p = 0.027) and LDL cholesterol (3.7%, p = 0.027). Conversely, shorter telomeres were associated with increased levels of IL-6 (7.2% decrease, p = 0.022). All analyses were adjusted for age, gender and deprivation group. The change in telomere length was measured by an increase of one standard deviation in log telomere length.
Adjustment covariates | % Change in IL6 Levels (95% CI) |
None | −9.9 (−16.3, −3.0), p = 0.006 |
Age + gender | −7.0 (−13.6, 0.1), p = 0.052 |
… + current smoking | −5.9 (−12.4, 1.0), p = 0.094 |
… + deprivation group | −5.7 (−11.9, 0.9), p = 0.090 |
… + income ≥/< £25,000 | −5.6 (−11.8, 1.0), p = 0.093 |
… + diet score ≥/< median | −5.6 (−11.8, 1.0), p = 0.096 |
This analysis was adjusted for the various different SES and lifestyle factors to determine their contribution to the difference in IL6 levels. The change in telomere length was measured by an increase of one standard deviation in log telomere length.
It has been hypothesized that socio-economic deprivation can accelerate biological ageing, resulting in shorter telomeres in deprived individuals in comparison to more affluent-aged matched controls. Five previous studies examining this relationship report positive
Interestingly, in the light of possible confounders relating to the veracity of SES data, employment status (men who reported being out of work) was reported to associate significantly with shorter telomeres
Our data are not incongrous with previous reports, as we observed no associations with area based deprivation and employment. However, we have demonstrated a direct link between accelerated biological ageing, low income and poor diet. Furthermore, we have observed a relationship with a measure of adiposity, namely waist/hip ratio (
These observations are intuitive and in keeping with the Marmot findings
Notably, these observations indicating an interaction between biological ageing and SES are reinforced by the finding that telomere length, in the pSoBid cohort, associates positively with LDL cholesterol levels, a strong and unambiguous causal risk factor in CVD.
In our study, telomere attrition was associated with increasing IL-6 levels, an emerging risk factor for CVD, which may predict fatal events more strongly than non-fatal events
Our findings suggest the unadjusted association between telomere length and IL-6 is strong, and it is still marginally significant with adjustment for age and gender. This association is only partially weakened by further adjusting for smoking but subsequent addition of deprivation, income and diet does not appear to weaken the IL-6-telomere association further (
Of the four previous studies in this field
The Q-PCR methodology employed in the present study yielded similar telomere length data in keeping with other reports
The difference in observations using telomere length as a marker of biological ageing, that have been reported by different groups have been elegantly summarised by Nordfjall et al
We differ in the detections of associations with total and LDL cholesterol, though these have previously been reported to be associated with telomere length in a disease setting
Our observations provide an intuitive link between proven socio-economic drivers of disease
This study may be limited by its size and cross sectional nature. Indeed, the social gradient in Glasgow is so extreme that a ‘survivor effect’ among the most deprived cannot be excluded. This merits a larger, longitudinal study to look at the relative impacts of further markers of SES and potential SE and lifestyle interventions. Such interventions are not without precedence and appear to show direct benefit to biological ageing. A recent intervention study in men with prostate cancer, reported that changing lifestyle, primarily via better diet and increased exercise leads to increased telomerase activity and deceleration of telomere attrition rate
In summary, we show convincingly that factors associated with lower socio-economic status and poor diet are relevant to accelerated biological ageing in a cohort representing extremes of social class. Our findings also suggest potential associations of elevated circulating IL-6, a measure known to predict CVD and diabetes, with biological ageing, observations which require further study to tease out potential mechanistic links.
The study was approved by the Glasgow Royal Infirmary Research Ethics Committee and all participants gave written informed consent.
The design of the psychological, social, and biological determinants of ill health (pSoBid) study has been described elsewhere
Measurement of biochemical parameters have been described in detail elsewhere
A diet score for the consumption of fruit and vegetables was calculated from subjects self-reported food frequency questionnaire responses. Participants were asked on average how often they consumed a range of food categories (21 food categories listed). Responses for each question ranged from daily consumption (number of portions per day) to weekly and monthly consumption. Participants selected one response per food category. For the purposes of the present analysis, responses to four questions from the food frequency questionnaire relating to fruit and vegetable intake were aggregated to give an overall indicative diet score (i.e. frequency of intake of fresh fruit, cooked green vegetables (fresh or frozen), cooked root vegetables (fresh or frozen) and raw vegetables or salad (including tomatoes)). Monthly diet scores were calculated on the basis of a 28 day month. The maximum possible total diet score was 672 (6 portions per day x 28 days per month x 4 food category questions).
DNA was extracted from PBLs following standard procedures and telomere lengths in the DNA samples were determined by Q-PCR, following the method of Cawthon
Relative telomere length was estimated from Ct scores using the comparative Ct method after confirming that the telomere and control gene assays yielded similar amplification efficiencies. This method determines the ratio of telomere repeat copy number to single copy gene number (T/S) ratio in experimental samples relative to a control sample DNA. This normalised T/S ratio was used as the estimate of relative telomere length (Relative T/S).
The inter-assay variation was assessed by comparing the relative telomere estimates (T/S ratio) across assays for the positive controls, which were assayed on every assay plate. The average inter-assay coefficient of variance was 0.3% for telomere and 0.1% for 36B4 plates.
Associations between telomere length and participant characteristics were investigated in linear regression models. Sampling was stratified by age, gender and deprivation group, and all models were adjusted for these factors. Because telomeres are expected to shorten gradually with age, age was represented in models as a continuous rather than categorical covariate.
When investigating factors that might influence ageing such as SES and lifestyle, telomere length was modelled as an outcome. Biomarkers, on the other hand, may be viewed as downstream of ageing, motivating their modelling as outcomes with telomere length as a covariate.
Telomere length and biomarkers were log-transformed for regression analysis to satisfy the assumption of normally distributed residuals. Regression coefficient estimates were therefore multiplicative when transformed back to the original scale. For example, a regression coefficient for a binary characteristic back-transormed to 1.1 implies the characteristic is associated with a 10% difference in the outcome. Thus, where log telomere length was the outcome, regression coefficients are presented as the percentage change in telomere length associated with each patient characteristic. Where telomere length was a covariate, a telomere length z-score was used so that the back-transformed regression coefficients could be interpreted as the percentage change in biomarker level associated with a one standard deviation increase in telomere length. The telomere length z-score was calculated by standardising log telomere length to have a mean of zero and standard deviation of one.
We hypothesised that the effects of telomere-shortening factors accumulate over time, therefore the largest differences between exposed and unexposed participants would be expected among the oldest participants. We investigated this hypothesis by testing for interactions between participant characteristics and age.
(DOC)
(DOC)