Pessimistic orientation in relation totelomere length in older men: The VANormative Aging Study

Ai Ikeda a, Joel Schwartz b, Junenette L. Peters c,Andrea A. Baccarelli b, Mirjam Hoxha d,e, Laura Dioni d,e,Avron Spiro f,g, David Sparrow h, Pantel Vokonas h,Laura D. Kubzansky a,*

aDepartment of Social and Behavioral Sciences, Harvard School of Public Health, Boston, MA, USAbDepartment of Environmental Health, Harvard School of Public Health, Boston, MA, USAcDepartment of Environmental Health, Boston University School of Public Health, Boston, MA, USAdCenter of Molecular and Genetic Epidemiology, IRCCS (Istituo Di Ricovero e Cura a Carattere Scientifico)Maggiore Hospital, Milan, ItalyeDepartment of Environmental and Occupational Health, University of Milan, Milan, ItalyfVA Normative Aging Study, VA Boston Healthcare System, Department of Epidemiology, Boston UniversitySchool of Public Health, Boston, MA, USAgDepartment of Psychiatry, Boston University School of Medicine, Boston, MA, USAhVA Normative Aging Study, VA Boston Healthcare System, Department of Medicine, Boston University Schoolof Medicine, Boston, MA, USA

Received 22 May 2013; received in revised form 25 December 2013; accepted 2 January 2014

Psychoneuroendocrinology (2014) 42, 68—76

KEYWORDSOptimism;Pessimism;Telomere length

Summary

Background: Recent research suggests pessimistic orientation is associated with shorter leuko-cyte telomere length (LTL). However, this is the first study to look not only at effects of pessimisticorientation on average LTL at multiple time points, but also at effects on the rate of change in LTLover time.Methods: Participants were older men from the VA Normative Aging Study (n = 490). The lifeorientation test (LOT) was used to measure optimistic and pessimistic orientations at studybaseline, and relative LTL by telomere to single copy gene ratio (T:S ratio) was obtainedrepeatedly over the course of the study (1999—2008). A total of 1010 observations were includedin the analysis. Linear mixed effect models with a random subject intercept were used toestimate associations.

* Corresponding author at: Department of Social and Behavioral Sciences, Harvard School of Public Health, 677 Huntington Avenue, Boston,MA 02115, USA. Tel.: +1 617 432 3589; fax: +1 617 432 3123.

E-mail address: lkubzans@hsph.harvard.edu (L.D. Kubzansky).

Available online at www.sciencedirect.com

ScienceDirect

jou rn a l home pag e : ww w. el sev ie r. com/ loca te /psyn eu en

0306-4530/$ — see front matter # 2014 Elsevier Ltd. All rights reserved.

http://dx.doi.org/10.1016/j.psyneuen.2014.01.001

Results: Higher pessimistic orientation scores were associated with shorter average LTL (percentdifference by 1-SD increase in pessimistic orientation (95% CI): �3.08 (�5.62, �0.46)), and thefinding was maintained after adjusting for the higher likelihood that healthier individuals return forfollow-up visits (�3.44 (�5.95, �0.86)). However, pessimistic orientation scores were not associatedwith rate of change in LTL over time. No associations were found between overall optimism andoptimistic orientation subscale scores and LTL.Conclusion: Higher pessimistic orientation scores were associated with shorter LTL in older men.While there was no evidence that pessimistic orientation was associated with rate of change in LTLover time, higher levels of pessimistic orientation were associated with shorter LTL at baseline andthis association persisted over time.# 2014 Elsevier Ltd. All rights reserved.

Pessimistic orientation in relation to telomere length in older men 69

1. Introduction

Telomeres are repetitive structures at the end of eukaryoticchromosomes which protect chromosome ends from deteriora-tion (Wong and Collins, 2001). Numerous studies have docu-mented the gradual shortening of telomere length withincreasing age (Slagboom et al., 1994). Additional work hasalso found shorter telomere length in patients suffering fromcoronary heart disease compared to controls (Brouilette et al.,2007), in diabetic patients (Sampson et al., 2006) and inindividuals with cardiovascular risk factors such as obesity,smoking (Nordfjall et al., 2008), as well as psychological stress(Epel et al., 2004). Thus, investigators have suggested thattelomere shortening is a marker of cellular aging that may beaccelerated by chronic stress and may indicate risk of chronicdisease development and premature mortality. A growing bodyof research has identified links between psychological factors(e.g., stress, pessimistic orientation, depression) and incidentdisease, disease progression, and mortality, but the mechan-isms underlying these associations remain to be determined.Building on the rapidly emerging science on telomeres, recentwork has begun to consider whether psychological functioningis associated with rate of cellular aging. For example, onecross-sectional study found higher levels of pessimistic orien-tation were associated with shorter telomere length (O’Dono-van et al., 2009).

Pessimism is a cognitive orientation (personality trait)characterized by the general expectation that outcomes willbe negative (Chang et al., 1997; Roy et al., 2010). High levelsof pessimistic orientation are associated with greater psy-chological distress; for example, pessimistic orientation is astrong predictor of depressive symptoms even after account-ing for other psychological factors (Chang et al., 1997). Morerecent work has found pessimistic orientation is also asso-ciated with physical health outcomes, with studies demon-strating higher levels of pessimistic orientation to beassociated with increased risk of heart disease and stroke,and reduced cancer survival (Kubzansky et al., 2001; Nabiet al., 2010; Novotny et al., 2010).

A number of mechanisms might explain the recentlyobserved association between pessimistic orientation and tel-omere shortening. Pessimistic orientation may induce negativereactions to stressful events thereby increasing the potentiallytoxic effects of stress-related neuroendocrine activation.Chronic neuroendocrine activation can increase oxidativestress, which induces telomeric DNA damage in the telomere

sequence (TTAGGG) (Kawanishi and Oikawa, 2004). TelomericDNA is synthesized by the enzyme telomerase. Thus, this DNAdamage may induce the slowing down of existing telomeraseenzymatic activity needed to synthesize telomoric DNA andthus accelerate telomere shortening (von Zglinicki, 2002).

Pessimistic orientation may also affect telomere shorteningthrough behavioral pathways such as increasing the likelihoodof smoking or being sedentary. These unhealthy behaviors havebeen demonstrated to have highly significant effects on bio-chemical processes (i.e., oxidative stress and inflammation)that can alter telomere dynamics (Nordfjall et al., 2008). Suchdamaging effects lead to shorter telomere length and morerapid age-dependent telomere attrition rate.

Only one study to date has examined LTL in relation topessimistic orientation (O’Donovan et al., 2009); other stu-dies have linked pessimistic orientation to factors associatedwith telomere length, including increased risk of inflamma-tion (O’Donovan et al., 2009), depression (Isaacowitz andSeligman, 2001), and premature mortality (Brummett et al.,2006). O’Donovan et al. (2009) suggested the associationwith LTL was stronger in relation to the pessimistic orienta-tion subscale as compared with either the optimistic orienta-tion subscale or the overall optimism score. This finding isconsistent with prior work we and others have done showingclearer effects of a pessimistic versus optimistic orientationon measures of inflammation and endothelial function as wellas other objective indicators of health (Milam et al., 2004;Ikeda et al., 2011). However, the initial study of pessimisticorientation and telomere length by O’Donovan et al. (2009),employed a cross-sectional design, and was conductedamong only 36 healthy post-menopausal women.

To gain greater insight into the relationship betweenpessimistic orientation and LTL and consider whether andhow pessimism may influence rate of LTL change, we examinethe association of a pessimistic or optimistic orientation withchange in telomere length over time in the VA NormativeAging Study (NAS). The life orientation test (LOT) was used tomeasure pessimistic or optimistic orientation in the presentstudy. In prior work, it has exhibited significant reliabilitywith a test—retest reliability reported to be 0.79 over a 4-week period and 0.69 over a 3-year period (Scheier andCarver, 1985), supporting the notion that dispositional opti-mism is relatively stable across time.

We considered a range of covariates which could be asso-ciated with the level of pessimistic or optimistic orientationscores in our sample including baseline (time at first visit when

(Recruited in 1960s.)2280

(Remaining in study by 1999.)1508

(With an LTL measure.)773

490 (Concurrent pessimism measures at the time of LTLmeasurement at visit one.)

(Additional LTL measure at visit two. )361

(Additional LTL measure at visit three.)154

(Additional LTL measure at visit four. )5

Figure 1 Flow chart of participants.

70 A. Ikeda et al.

optimistic/pessimistic orientation and LTL were measured)age, the difference in age between baseline and time at whicheach LTL was measured, body mass index (BMI), educationattainment level, cigarette smoking status, alcohol drinking,physical activity, medical conditions (i.e., hypertension, dia-betes, cardiovascular disease) and interleukin-6 (IL-6). Allcovariates were chosen based on prior literature suggestingtheir potential relevance to both telomere change and pessi-mistic or optimistic orientations (Nilsson, 2011; Needhamet al., 2013). We further considered possible effect modifica-tion by age because age at study baseline could be a strongfactor associated with LTL while pessimism is not necessarystrongly associated with age. Moreover, other studies havesuggested the effects of psychological factors on health-related outcomes are often stronger at younger versus olderages (Hamer et al., 2008). Thus, we believed that age mightalter finding not because the trait varies by age, but becausethe effects of trait may vary by age.

2. Methods

2.1. Study population

The VA Normative Aging Study (NAS) has been described indetail elsewhere (LoCastro et al., 2000). In brief, the NAS is alongitudinal study of aging established by the Veterans Admin-istration (now the Department of Veterans Affairs) between1961 and 1963 among 2280 men (98% Caucasian, 1% African-American, and 1% Hispanic) from the Greater Boston area aged21—80 years at the time of entry, who were free of any knownchronic medical conditions. Participants are community-resid-ing men, most of whom were veterans. Many veterans hadcombat experience but not all; it was not a requirement for thestudy. Men were asked to return for onsite physical examina-tions and questionnaires every 3—5 years. Study participantsprovided written informed consent and the study protocol wasapproved by the Institutional Review Boards of all participatinginstitutions. Eligibility for the present study required contin-ued participation as of the time when telomere measurementbegan (1999). Drop out has been less than 1% per year in thiscohort, and predominantly occurs only if participants move outof the study area. The other major reason for loss to follow-uphas been mortality.

Participants come to the clinic for each physical examina-tion and blood is drawn at each visit. Telomere length wasmeasured at each exam between 1999 and 2008 (the presentstudy period). Participants were also administered a series ofquestionnaires that included items measuring an optimisticor pessimistic orientation and other covariates during thepresent study period. The present study included a total of490 men who were active participants, provided at least oncetelomere measure, and concurrently responded to all opti-mistic and pessimistic orientation items at the time when theleukocyte telomere length (LTL) was first measured (Fig. 1).Of these, 5 (1.0%) had 4 telomere length measurements, 149(30.4%) had 3 telomere length measurements, 207 (42.2%)had 2 telomere length measurements, and 129 (26.3%) hadone telomere length measurement. Thus, a total of 1010observations were used in the present analysis.

To assess whether subjects excluded from our study weredifferent from those included, we compared some baselinecharacteristics such as age, BMI, smoking status, drinking

behavior, physical activity, educational level, and health con-ditions (hypertension, diabetes, and cardiovascular disease)between our included subjects and excluded subjects (n = 327)who came in during the study period (1998—2008) but did nothave LTL measures and/or valid responses to the optimistic andpessimistic orientation measures. There were no differences inthe characteristics between excluded and included subjects.

2.2. Assessment of subscales for optimisticorientation and pessimistic orientation

The LOT was used to measure optimistic and pessimisticorientations (Ikeda et al., 2011). The LOT is a validatedinstrument (Scheier and Carver, 1985) made up of 8 relevantitems comprising questions such as ‘‘In unclear times, Iusually expect the best’’; ‘‘If something can go wrong forme, it will.’’ We tested reliability of each scale or subscale(i.e., overall optimism, subscales of optimism and pessimism)by using Cronbach’s a (alpha), a coefficient that assesses theinternal consistency of the items (intercorrelation within theitems). Cronbach alpha coefficient was 0.81 for overall; 0.78for optimistic orientation subscale; 0.82 for pessimistic sub-scale. In the analysis, we used the concurrent measurementsfor optimistic and pessimistic orientations taken at the timewhen the LTL was first measured.

2.3. Telomere length measurements

We collected 7 mL of whole blood by venous phlebotomy inEDTA tubes. DNA was extracted from stored frozen buffy coatusing the QiAmp DNA blood kits (Qiagen, Germantown, MD,USA) and used to obtain LTL measurement by means ofquantitative real-time polymerase chain reaction (qRT-PCR) (Cawthon, 2002). The assay process is described indetail elsewhere (McCracken et al., 2010). In brief, relativeLTL was measured on a 7900HT Fast Real-Time PCR System(Applied Biosystems, Foster City, CA, USA) by determining theratio of the telomere (T) repeat copy number to the single-copy gene (S) copy number (T:S ratio) in a given sample. LTLwas reported as relative units expressing the ratio between

Pessimistic orientation in relation to telomere length in older men 71

LTL in the test DNA and LTL in a reference DNA pool. We ran allsamples in triplicates, and the average of the three T mea-surements was divided by the average of the three S mea-surements to calculate the average T:S ratio (Farzaneh-Faret al., 2008). LTL measures were assayed all at once; there-fore we did not adjust for potential batch effects.

2.4. Measurement of other covariates

Every 3—5 years, participants in the NAS are followed up byphysical examination, updating of medical history, height,weight, measurement of biomarkers including fasting glucose,and complete a questionnaire including various measures oflifestyle and psychosocial factors (these may vary acrossexams). Educational attainment level (years), cigarette smok-ing status (current, former, or never) and alcohol drinking (lessthan two versus two or more drinks per day) are ascertained bya trained interviewer. Current smokers (yes, no) are defined asmen who smoke 1 or more cigarettes per day. Responses toquestions about the number of flights of stairs climbed per day,walking pace, and frequency of various sports activities wereused to derive a continuous physical activity variable thatassessed total kilocalories expended per week (METS-hr/wk).Interleukin-6 (IL-6) was assayed with the Milliplex Map HumanCytokine/Chemokine Kit (Millipore Corporation, MO, USA) anddetected with a multiplex detection platform (Luminex1 100/200TM System, Austin, TX, USA).

2.5. Statistical analysis

Analysis of covariance and chi-square tests were used tocompare mean values and proportions of covariates. LTL wasnatural log-transformed for the analyses to improve the nor-mality of the data distribution. The present study data con-sisted of LTL measurements repeated over time, meaning thatwe have longitudinal data with multiple measurements on eachindividual obtained over time. Analysis of these repeatedmeasures were performed using mixed effects modeling, thecommonly accepted method for dealing with longitudinal data,which accounts for the correlation existing between measure-ments taken from the same individual (Cnaan et al., 1997). Inthe present study, we used the following model:

yi j ¼ ðb0 þ b0iÞ þ bP � Pessimismi þ bT � Timei j þ bPT

� ðPessimism � TimeÞi j þ ei j

where yij represents the logarithm of the telomere length forindividual i taken at time j; b0 and bT represent respectivelythe intercept and the slope of the linear relationshipbetween the logarithm of the LTL and time; bP is the effectof a unit increase in pessimistic orientation on log (telomerelength), considered as constant across time (what we called a‘‘pooled’’ across time); bPT is the effect of a unit increase inpessimistic orientation on the slope describing the linearrelationship between log (LTL) and time. Coefficients for thismodel were estimated by maximization of the likelihoodusing the SAS procedure MIXED and specifying a compoundsymmetry structure for the covariance matrix. We alsoreported LTL means at each time point according to thelevels of pessimism (low vs. medium to high) among subjectswho had at least three sets of visits (n = 154). For interpret-

ability, exponentiated regression coefficients from regressionmodels with logged continuous outcomes were reported aspercent differences in LTL associated with a 1-standarddeviation increase in pessimistic orientation and their 95%confidence intervals (95% CI) after adjustment for age andother potential confounding factors. We also calculated andpresented the percent difference in LTL according to quar-tiles of pessimistic orientation to assess the possibility of athreshold effect on telomere length.

In multivariate analysis, we only adjusted for covariatesthat were associated with the level of pessimistic orientationscore at a significant level of p < 0.10 in our sample, includ-ing baseline age (years), delta age (the difference in agebetween baseline and time at which the outcome was mea-sured), body mass index (kg/m2), glucose category (normal,borderline or diabetes mellitus), education attainment level(<12, �12 years). Borderline diabetes mellitus was definedas a fasting glucose level of 6.1—6.9 mmol/l. Diabetes mel-litus was defined as a fasting glucose level of �7.0 mmol/l orthe use of medication for diabetes. Covariate measurementswere assessed at baseline of the study period, which was alsowhen pessimistic orientation was assessed, but if measure-ments were missing we used available measurements closestin time when pessimistic orientation was assessed. To assessthe potential modifying effects of baseline age, and changein age on the relation of pessimistic orientation with LTL overtime, we ran regression models that included a cross-productterm for interaction between age and change in age with thepessimistic orientation score along with the main-effects.

To adjust for the possibility that healthier men are morelikely to return for subsequent exams (revisit), a propensityscore was used to model the probability of multiple exams,and inverse probability weights were used to correct forpossible selection bias in the analyses using repeated mea-sures over time. This revisit propensity score was calculatedfrom a logistic regression as the probability of not having twoor more study center visits, given all relevant factors atbaseline including age; measurement of telomere length;total cholesterol level; statin use; abnormal glucose toler-ance; hypertension; body mass index; smoking status; alcoholuse, education and attainment. Thus, the primary analysiswas repeated after weighting follow-up observations by theinverse probabilities of attaining follow-up response (revisit)(Kurth et al., 2005).

To further address possible concerns about whether otherfactors influencing likelihood of revisits might also influenceassociation with LTL, we conducted a sensitivity analysisremoving subjects who only had one measurement of LTLand the primary analysis was repeated within the remainingparticipants. Moreover, to address the issue of LTL lengthen-ing, a function of ‘‘noise’’ in the measurement process, weran an additional analysis excluding subjects whose LTLlengthened from baseline and reanalyzed effects withinthe remaining participants (n = 584).

All analyses were conducted using the SAS statisticalpackage Version 9.1 (SAS Institute Inc., Cary, NC, USA).

3. Results

Characteristics of the study sample are presented in Table 1according to quartiles of pessimistic orientation score. At

Table 1 Distribution of baseline characteristics in 490 men according to the quartiles of pessimistic orientation subscale score.

Range Level of pessimistic orientation subscale P-value

Lowest Highest�3 4 5—6 7+

No. at risk 116 135 127 112Visits (SD) 1.72 (0.76) 1.70 (0.77) 1.65 (0.73) 1.63 (0.71) 0.15Baseline age, year (SD) 71.9 (0.62) 71.3 (0.57) 71.9 (0.59) 73.1 (0.63) 0.22BMI, kg/m2 (SD) 27.4 (0.38) 28.0 (0.36) 28.5 (0.37) 28.8 (0.39) 0.06Hypertension, % 62.9 58.5 55.9 68.8 0.19Diabetes, % 11.2 12.6 18.1 22.3 0.07Physical activity, METS-hr/wk (SD) 13.7 (1.7) 16.8 (1.6) 13.6 (1.7) 11.7 (1.8) 0.20History of cardiovascular disease, % 31.0 25.2 23.6 33.0 0.30Current smoker, % 5.17 3.70 1.57 8.04 0.11Two or more drinks, % 18.1 18.5 17.3 17.9 0.99Educational attainment level, year (SD) 15.2 (0.27) 14.4 (0.25) 14.1 (0.25) 14.1 (0.27) <0.001

72 A. Ikeda et al.

baseline, men with higher pessimistic orientation scores weremore likely to be current smokers and have higher BMI andlower levels of educational attainment. In addition, more menwith high levels of pessimistic orientation had diabetes sug-gesting that pessimistic orientation is initially associated withworse health. The mean age of the study population was 72.0years at baseline (SD = 6.7 years; median = 71 years; range 55—90 years). The mean pessimistic orientation score was 4.7(SD = 2.6) and was normally distributed in this sample. Theaverage LTL for subjects who only had one visit was 1.17(SD = 1.41) and for those who had more than one visit was1.17 (SD = 1.41). The average follow-up time of participantswho had at least two visits was 3.5 (SD = 0.97, range = 2—8) upto visit 2, 6.25 (SD = 0.70, range = 4—9) up to visit 3 and 8.00(SD = 0.70, range = 7—9) up to visit 4.

Age-adjusted and multivariable-adjusted percent differ-ences in LTL (pooled across all time points) associated with a1-SD difference in pessimistic orientation subscale score andwith quartiles of pessimistic orientation subscale score (low-est as a reference category) are reported in Table 2. In theage-adjusted model higher levels of pessimistic orientationwere associated with shorter LTL (main-effect) and this

Table 2 Age- and multivariate-adjusted percent difference assoc(the lowest category as reference) or with a 1-standard deviation intransformed telomere length.

Level of pessimism subscale

Low

�3 4 5—6

Crude 0 �3.79 (�10.5, 3.44) �5.63 (�Age-adjusteda 0 �4.17 (�10.9, 3.07) �5.96 (�Multivariate-adjusted 0 �3.93 (�10.7, 3.37) �5.73 (�Further weighted forinverse probabilityof revisits b

0 �3.79 (�10.6,3.49) �5.86 (�

In multivariate-adjusted model, adjusted for baseline age, body massa Adjusted for age at baseline and ^age (difference in age between

b The revisit propensity score was calculated from a logistic regressiogiven all relevant factors at baseline including age; telomere lengthhypertension; alcohol intake; smoking status; and education attainme

association was maintained after adjusting for all covariates.Thus, a 1-SD increase in pessimistic orientation was asso-ciated with a 3.1% decrease in LTL. The association becamesomewhat stronger after including the revisit propensityscore such that a 1-SD increase in pessimistic orientationwas associated with a 3.4% decrease in LTL. When we furtheradjusted for baseline IL-6, the associations were not atte-nuated. The percent difference by 1-SD increase in pessi-mistic orientation after adjusting for IL-6 was for IL-6 was�3.44 (95% CI: �5.96, �0.85). We also found that individualswith higher pessimistic orientation scores had shorter LTLcompared to those in lowest pessimistic orientation scorecategory. A graded relationship was also found between LTLand the ordinal category of pessimistic orientation, wherebywith each increase in level of pessimistic orientation, LTL wasshorter. Neither the overall optimism nor the optimisticorientation subscale were associated with LTL in any analysisof our sample (Appendix A). In sensitivity analysis removingparticipants who had only one measurement of LTL, resultswere unchanged in the final model which included the revisitpropensity score (percent LTL difference by 1-SD increase inpessimistic orientation, �3.63 (95% CI: �6.46, �0.71)).

iated with the quartile of pessimistic orientation subscale scorecrease in pessimistic orientation subscale score (95% CI) for log-

High 1SD change7+ 1SD = 2.605

12.4, 1.62) �8.26 (�15.0, �0.96) �3.20 (�5.68, �0.65)12.7, 1.28) �8.42 (�15.2, �1.10) �3.18 (�5.68, �0.61)12.6, 1.63) �8.17 (�15.1, �0.70) �3.08 (�5.62, �0.46)12.7,1.54) �9.04 (�15.8,�1.70) �3.44 (�5.95,�0.86)

index, abnormal glucose tolerance, education, and ^age.baseline and time of outcome).n as the probability of not having two or more study center visits,; total cholesterol level; statin use; abnormal glucose tolerance;nt level.

Figure 2 Analysis of covariance was used to obtain meanvalues of telomere length (SD) according to two levels of pessi-mism (low (�3) vs. medium to high (4+)). We used 154 subjectswho had at least 3 sets of visits.

Pessimistic orientation in relation to telomere length in older men 73

An overall effect of time in the present study was sig-nificantly associated with the loss of telomere length (per-cent difference in LTL associated with a difference in agebetween baseline and time (one year increment) when LTLmeasured: �1.94 (95% CI: �2.91, �0.96). However, no asso-ciation of pessimistic orientation with change in LTL overtime was evident. In addition, no statistically significantinteractions between pessimistic orientation and baselineage or delta age were evident for either LTL pooled acrosstime or change in LTL over time. For example, the multi-variable-adjusted (including inverse probability of multipleexams) percent difference (95% CI) for the interactionbetween a 1-SD in pessimistic orientation and baseline ageor delta age in relation to LTL was 0.10 (�0.28, 0.47) and�0.25 (�1.25, 0.76), respectively. While there was no evi-dence that pessimistic orientation was associated with rateof change in LTL over time, we found some LTL mean differ-ences by the levels of pessimism at each time point. The LTLmean was lower among subjects with medium to high level ofpessimistic orientation at baseline and the differences per-sisted over time; however, these differences were not sta-tistically significant (Fig. 2).

3.1. Additional analyses

After we excluded the subjects who had LTL lengthened sincebaseline, the effects of pessimism on LTL became weaker andmarginally significant likely due to the loss of statisticalpower, but the tendency of the effect was still evidentand consistent (percent difference by 1-SD increase in pessi-mistic orientation, �2.95 (95% CI: �6.02, 0.23; p = 0.07)). Wealso compared baseline characteristics between subjectswho had LTL lengthened since baseline and those who didnot. There were differences in mean LTL (1.35 vs. 1.04) andbaseline age (73 vs. 71 years), but no other differences incharacteristics (i.e., BMI, history of hypertension, diabetes,and cardiovascular disease, smoking status, drinking beha-vior, physical activity, and educational level).

4. Discussion

In this analysis with repeated measures of telomere length,we found a consistent association between pessimistic orien-

tation and LTL pooled over multiple measures across time inelderly men, but no relation with LTL rate of change overtime. Findings were maintained after adjusting for potentialconfounders and little attenuation of effect was evident evenafter adjusting for behaviors that could be on the pathwaybetween pessimistic orientation and accelerated aging. Wefound no evidence of associations between measures ofoverall optimism or optimistic orientation and either pooledmeasures of telomere length or rate of change in the telo-mere length over time.

Processes that influence the rate of cell replication andthe extent of telomere loss during each replication willinfluence how rapidly telomeres shorten. Chronic inflamma-tion, for example, is associated with an increase in theturnover rate of leukocytes, which may lead to the higherrate of replication of cells. We have previously reported thatpessimistic orientation was associated with higher levels ofinflammation and endothelial function (Ikeda et al., 2011).Including IL-6 in our model did not significantly alter ourfindings. However, a full test of mediation requires additionallongitudinal models beyond the scope of this present studybut may be an important direction for future work.

Our findings were also strongly consistent with that of theprior study, which demonstrated a cross-sectional associationbetween higher pessimistic orientation and shorter LTLamong 36 post-menopausal women; optimistic orientationwas also not associated with LTL in this study (O’Donovanet al., 2009). Unlike this study, we examined the effect ofpessimistic orientation on multiple measures of LTL pooledacross time, and also on change in LTL over time. Poolingacross multiple measures of LTL over time likely reducesnoise that may be present among measures taken at a singlepoint in time, and results in a stronger measure of the truelevel of the parameter being assessed.

However, we found no relation between pessimistic orien-tation and the rate of change in the LTL over time. Onepossible reason for this is the wide temporal variability inpatterning of telomere length changes over time (Farzaneh-Far et al., 2010). However, we also observed increases in LTLbetween measurements among some participants ratherthan the monotonic decrease in length expected to occurwith aging. For example, we found increases in LTL among32% of subjects (n = 155) between baseline and 2nd measure-ment, and continuing increase in LTL among 10% of them atthe 3rd measurement. Sensitivity analyses excluding parti-cipants whose LTLs lengthened or comparing these partici-pants with those whose LTLs shortened did not significantlychange the interpretation of our findings. Furthermore,despite the small coefficient of variation (8.7%) in the pre-sent study among the triplicate measures of the T:S ratios,even relatively low measurement error can result in mis-classification (Chen et al., 2011) The appearance of LTLlengthening can be more a function of ‘‘noise’’ in the mea-surement process than a true effect (Chen et al., 2011). Inaddition, PCR-based measures are considered somewhatmore susceptible to age-related changes in DNA becausetelomere length is expressed in relation to a reference genewhich is presumed to be stable (Aviv et al., 2009). The olderage of our cohort at baseline, relatively short follow-upperiod, and PCR-based measures of telomeres may havelimited our ability to assess accurately change in telomerelength over time in this cohort.

74 A. Ikeda et al.

To the best of our knowledge, the present study is thefirst to test the hypothesis that pessimistic orientationleads to changes in LTL by looking at the relationshipbetween pessimistic orientation and repeated measuresof LTL over time while adjusting for a range of potentialconfounders or possible behavioral pathways. Our findingsof no association of pessimism with rate of change in LTLover time but a constant effect of unit increase across timeare consistent with the explanation that rate of changetrajectories are set earlier in life at younger ages or at anearlier stage with regard to telomere dynamics. If that isthe case, then maintenance of relative differences in LTLwill be evident at older ages, but differences in rates maynot be.

We failed to find the associations with optimistic orienta-tion. Previous findings (O’Donovan et al., 2009; Roy et al.,2010; Ikeda et al., 2011) have suggested that physiologicaleffects are more strongly evident with a pessimistic versusoptimistic orientation. It is possible that this is due to methodartifact, for example, negatively worded items (i.e., pessi-mistic orientation items) may be answered differently frompositively worded items particularly in older adult popula-tions. This may result in differential associations with out-comes and optimistic orientation may be less related in oldermen than in younger people (Mroczek et al., 1993); this mayhelp to explain why we did not find associations between LTLand optimistic orientation or overall optimism. Moreover, LTLprovides only one measure of telomere dynamics. In thepresent study, we were not able to measure telomeraseexpression and activity, and thus, could not examine whethertelomerase levels were associated with pessimistic or opti-mistic orientations. Telomerase, is a cellular enzyme whichadds telomeric repeat sequences to the telomere regions atthe ends of eukaryotic chromosomes (Aviv et al., 2011).The sequence of elongation by telomerase is one of theprocesses maintaining telomeric intactness and healthycell function (Aviv et al., 2011; Hohensinner et al.,2011). Other work has suggested that telomerase activitymay provide additional insight into telomere dynamicsoffering some protective effects and thereby perhaps morestrongly associated with more positive attributes like anoptimistic orientation. Thus, telomerase levels, in whichshort term changes are measurable (Weng et al., 1996),may provide additional insight into telomere dynamics andperhaps also provide an earlier or more sensitive indicatorthan LTL of cell aging and of disease processes (e.g.,cardiovascular disease) (Epel et al., 2006). Additional workwith multiple measures of telomere length as well astelomerase obtained over time will help to clarify howthese dynamics are related to both upstream factors likepessimistic orientation as well as subsequent health (Avivet al., 2009; Martin-Ruiz et al., 2005).

We note several other limitations of our study. First, theratio of Tand S measurements we used in the present study isoften used to represent absolute telomere length; however,this measurement only provides the average telomere lengthper sample. Second, the observed associations in the presentstudy may be biased toward null since this study was limitedto a cohort of elderly who may have relatively long telomeresfrom early in life (Heidinger et al., 2012). The subjects werenot all veterans although most were. Thus, our findings willneed to be replicated in other population to assess general-

izability. That said, to generalize research findings, the NASstudy was originally designed to cover as wide a range aspossible of socioeconomic variables (Bell et al., 1972). Third,as in most longitudinal studies, there was imbalance in thenumber of repeated measures per subject, here largely dueto loss to follow-up (i.e., poor baseline health condition).However, concerns about how this might affect the findingsare mitigated given the similar results obtained after usinginverse probability of response weighting. Fourth, we alsonote that our findings are limited to men but were consistentwith findings from the prior study in women (O’Donovanet al., 2009). Finally, although we adjusted for variouspossible confounding factors in the current study, there isthe possibility of residual confounding by unmeasured vari-ables such as external stressor (i.e., caregiving or loss ofspouse). External stressors might modify the effect of pessi-mism on telomeres or pessimism might mediate the effect ofexternal stressors on telomere length. We have previouslyfound that both adulthood and childhood socioeconomicstatus (SES), one form of external stressor, were each inde-pendently associated with pessimism in this sample (Peterset al., 2011). Childhood SES may effect on the developmentalorigin of pessimistic orientation and have lasting effects intoadulthood. Socioeconomic achievement during adulthoodmay also be associated with stability in pessimistic orienta-tion (Ek et al., 2004). This may suggested that pessimism canmediate effects of SES on LTL. Our adjustment for education(a marker of SES) as a potential confounder in the multi-variate model did not significantly altered findings and more-over education was not directly associated with LTL (data notshown). A recent study found that less education (less thanhigh school) compared to more education (college degree)was inversely associated with LTL (Needham et al., 2013).Compared to the study conducted by Needham et al., ourstudy sample was much older (mean baseline age was 72.0(SD = 6.7)); therefore, any effects of having lower educationmay have happened, but no longer be apparent during ourstudy period. Furthermore, the proportion of less than highschool was smaller (6%) compared to the previous study (16%for white population and 32% for full sample) (Needhamet al., 2013). Due to small variation in the presented sample,we may not be able to detect the effect of lower education onLTL in our present study.

Strengths of the study include the use of a well-charac-terized cohort of elderly men who have been followed overtime among whom attrition is low. Pessimistic orientationwas measured with a widely used, validated measure (Kub-zansky et al., 2004; Scheier et al., 1994). Other work hasindicated that pessimistic orientation is a stable trait andremains relatively unchanging over the course of adulthoodand even after receipt of bad news (Schou et al., 2005). Allmeasures were obtained at the baseline visit and at least twoor more measures of LTL were available for most subjects(73.6%).

In summary, we found that a more pessimistic orientationwas associated with lower LTL in community-dwelling oldermen when pooled across multiple points in time. The asso-ciation was maintained after adjusting for a standard set ofbiological and behavioral factors that could potentially con-found or mediate the association of pessimistic orientationwith LTL, suggesting that other mechanisms should be con-sidered. Efforts at prevention and intervention for healthy

Pessimistic orientation in relation to telomere length in older men 75

aging may benefit from a broader focus on psychologicalresources that are associated with biological wear and tear.

Role of funding source

This study was supported by the Robert Wood Johnson Foun-dation’s Pioneer Portfolio,through a grant, ‘‘Exploring Con-cepts of Positive Health’’, NIEHS 2RO1-ES015172, and NIHAG018436. The VA Normative Aging Study is supported by theCooperative Studies Program/ERIC, US Department of Veter-ans Affairs, and is a research component of the MassachusettsVeterans Epidemiology Research and Information Center(MAVERIC). This study was also supported by a Research

Appendix A

Age- and multivariate-adjusted percent difference associateorientation subscale score (the lowest category as reference) andlog-transformed telomere length.

Level of overall optimism

Low

�19 20—22 23—

Crude 0 �0.73 (�7.48, 6.52) 1.18Age-adjusteda 0 �0.76 (�7.52, 6.51) 1.32Multivariate-adjusted 0 �1.27 (�8.11, 6.08) 1.18Further weighted forinverse probabilityof revisitsb

0 �0.60 (�7.50, 6.82) 2.49

Level of optimistic orientation subscale

Low

�9 10 11—12

Crude 0 �1.91 (�11.3, 0.37) �4.44 (Age-adjusteda 0 �3.69 (�11.0, 4.23) �4.22 (Multivariate-adjusted 0 �4.17 (�11.5, 3.79) �4.55 (Further weighted forinverse probabilityof revisitsb

0 �3.51 (�10.9, 4.55) �3.70 (

In multivariate-adjusted model, adjusted for baseline age, body mass ina Adjusted for age at baseline and ^age (difference in age between bab The revisit propensity score was calculated from a logistic regression as

relevant factors at baseline including age; telomere length; total cholealcohol intake; smoking status; and education attainment level.

Career Scientist awards for VA Clinical Science R&D Serviceto Avron Spiro, III and to David Sparrow.

Conflict of interest

There are no conflicts of interests.

Acknowledgement

The authors sincerely express their appreciation to Dr.Hadrien Charvat, National Cancer Center, Tokyo, for valuablecomments on the manuscript.

d with the level of overall optimism score and optimistic a 1-standard deviation increase in optimism score (95% CI) for

High 1SD change24 25+ 1SD = 4.249

(�6.25, 9.22) 3.96 (�3.69, 12.2) 1.37 (�1.25, 4.05) (�6.15, 9.39) 4.06 (�3.64, 12.4) 1.39 (�1.24, 4.09) (�6.38, 9.35) 3.48 (�4.31, 11.9) 1.26 (�1.40, 4.00) (�5.08, 10.7) 4.61 (�3.27, 13.1) 1.82 (�0.84, 4.56)

High 1SD change13+ 1SD = 2.496

�10.5, 1.99) �1.91 (�9.62, 6.46) �1.08 (�3.63, 1.55)�10.3, 2.26) �1.73 (�9.49, 6.69) �1.01 (�3.57, 1.62)�10.7, 1.99) �2.09 (�9.89, 6.39) �1.09 (�3.67, 1.57)�9.86, 2.88) �1.18 (�9.05, 7.36) �0.57 (�3.16, 2.08)

dex, abnormal glucose tolerance, education, and ^age.seline and time of outcome).

the probability of not having two or more study center visits, given allsterol level; statin use; abnormal glucose tolerance; hypertension;

References

Aviv, A., Chen, W., Gardner, J.P., Kimura, M., Brimacombe, M., Cao,X., Srinivasan, S.R., Berenson, G.S., 2009. Leukocyte telomeredynamics: longitudinal findings among young adults in the Boga-lusa Heart Study. Am. J. Epidemiol. 169, 323—329.

Aviv, A., Hunt, S.C., Lin, J., Cao, X., Kimura, M., Blackburn, E., 2011.Impartial comparative analysis of measurement of leukocytetelomere length/DNA content by Southern blots and qPCR.Nucleic Acids Res. 39, e134.

Bell, B., Rose, C., Damon, A., 1972. The Normative Aging Study: aninterdisciplinary and longitudinal study of health and aging. AgingHum. Dev. 3, 4—17.

Brummett, B.H., Helms, M.J., Dahlstrom, W.G., Siegler, I.C., 2006.Prediction of all-cause mortality by the Minnesota MultiphasicPersonality Inventory Optimism-Pessimism Scale scores: study ofa college sample during a 40-year follow-up period. Mayo Clin.Proc. 81, 1541—1544.

Brouilette, S.W., Moore, J.S., McMahon, A.D., Thompson, J.R., Ford,I., Shepherd, J., Packard, C.J., Samani, N.J., West of ScotlandCoronary Prevention Study Group, 2007. Telomere length, risk ofcoronary heart disease, and statin treatment in the West ofScotland Primary Prevention Study: a nested case—control study.Lancet 369, 107—114.

Cawthon, R.M., 2002. Telomere measurement by quantitative PCR.Nucleic Acids Res. 30, e47.

76 A. Ikeda et al.

Chang, E.C., Maydeu-Olivares, A., D’Zurilla, T.J., 1997. Optimismand pessimism as partially independent constructs: relationshipto positive and negative affectivity and psychological well-being.Pers. Individ. Diff. 23, 433—440.

Chen, W., Kimura, M., Kim, S., Cao, X., Srinivasan, S.R., Berenson,G.S., Kark, J.D., Aviv, A., 2011. Longitudinal versus cross-sec-tional evaluations of leukocyte telomere length dynamics: age-dependent telomere shortening is the rule. J. Gerontol. A: Biol.Sci. Med. Sci. 66, 312—319.

Cnaan, A., Laird, N.M., Slasor, P., 1997. Using the general linearmixed model to analyze unbalanced repeated measures andlongitudinal data. Stat. Med. 16, 2349—2380.

Ek, E., Remes, J., Sovio, U., 2004. Social and developmental pre-dictors of optimism from infancy to early adulthood. Soc. Indic.Res. 69, 219—242.

Epel, E.S., Blackburn, E.H., Lin, J., Dhabhar, F.S., Adler, N.E.,Morrow, J.D., Cawthon, R.M., 2004. Accelerated telomere short-ening in response to life stress. Proc. Natl. Acad. Sci. U.S.A. 101,17312—17315.

Epel, E.S., Lin, J., Wilhelm, F.H., Wolkowitz, O.M., Cawthon, R.,Adler, N.E., Dolbier, C., Mendes, W.B., Blackburn, E.H., 2006. Cellaging in relation to stress arousal and cardiovascular disease riskfactors. Psychoneuroendocrinology 31, 277—287.

Farzaneh-Far, R., Cawthon, R.M., Na, B., Browner, W.S., Schiller,N.B., Whooley, M.A., 2008. Prognostic value of leukocyte telo-mere length in patients with stable coronary artery disease: datafrom the Heart and Soul Study. Arterioscler. Thromb. Vasc. Biol.28, 1379—1384.

Farzaneh-Far, R., Lin, J., Epel, E., Lapham, K., Blackburn, E.,Whooley, M.A., 2010. Telomere length trajectory and its deter-minants in persons with coronary artery disease: longitudinalfindings from the heart and soul study. PLoS ONE 5, e8612.

Hamer, M., Molloy, G.J., Stamatakis, E., 2008. Psychological dis-tress as a risk factor for cardiovascular events: pathophysiolo-gical and behavioral mechanisms. J. Am. Coll. Cardiol. 52,2156—2162.

Heidinger, B.J., Blount, J.D., Boner, W., Griffiths, K., Metcalfe, N.B.,Monaghan, P., 2012. Telomere length in early life predicts life-span. Proc. Natl. Acad. Sci. U.S.A. 109, 1743—1748.

Hohensinner, P.J., Goronzy, J.J., Weyand, C.M., 2011. Telomeredysfunction, autoimmunity and aging. Aging Dis. 2, 524—537.

Ikeda, A., Schwartz, J., Peters, J.L., Fang, S., Spiro 3rd., A., Sparrow,D., VokonasF P., Kubzansky, L.D., 2011. Optimism in relation toinflammation and endothelial dysfunction in older men: the VANormative Aging Study. Psychosom. Med. 73, 664—671.

Isaacowitz, D.M., Seligman, M.E., 2001. Is pessimism a risk factor fordepressive mood among community-dwelling older adults?Behav. Res. Ther. 39, 255—272.

Kawanishi, S., Oikawa, S., 2004. Mechanism of telomere shorteningby oxidative stress. Ann. N. Y. Acad. Sci. 1019, 278—284.

Kubzansky, L.D., Sparrow, D., Vokonas, P., Kawachi, I., 2001. Is theglass half empty or half full? A prospective study of optimism andcoronary heart disease in the normative aging study. Psychosom.Med. 63, 910—916.

Kubzansky, L.D., Kubzansky, P.E., Maselko, J., 2004. Optimism andpessimism in the context of health: bipolar opposites or separateconstructs? Pers. Soc. Psychol. Bull. 30, 943—956.

Kurth, T., Walker, A.M., Glynn, R.J., Chan, K.A., Gaziano, J.M.,Berger, K., Robins, J.M., 2005. Results of multivariable logisticregression, propensity matching, propensity adjustment, andpropensity-based weighting under conditions of nonuniformeffect. Am. J. Epidemiol. 163, 262—270.

LoCastro, J., Spiro 3rd., A., Monnelly, E., Ciraulo, D., 2000. Person-ality, family history, and alcohol use among older men: the VANormative Aging Study. Alcohol. Clin. Exp. Res. 24, 501—511.

Martin-Ruiz, C.M., Gussekloo, J., van Heemst, D., von Zglinicki, T.,Westendorp, R.G., 2005. Telomere length in white blood cells is

not associated with morbidity or mortality in the oldest old: apopulation-based study. Aging Cell 4, 287—290.

McCracken, J., Baccarelli, A., Hoxha, M., Dioni, L., Melly, S., Coull,B., Suh, H., Vokonas, P., Schwartz, J., 2010. Annual ambient blackcarbon associated with shorter telomeres in elderly men: Veter-ans Affairs Normative Aging Study. Environ. Health Perspect. 118,1564—1570.

Milam, J.E., Richardson, J.L., Marks, G., Kemper, C.A., McCutchan,A.J., 2004. The roles of dispositional optimism and pessimism inHIV disease progression. Psychol. Health 19, 167—181.

Mroczek, D.K., Spiro III, A., Aldwin, C.M., Ozer, D.J., Bosse, R., 1993.Construct validation of optimism and pessimism in older men:findings from the normative aging study. Health Psychol. 12, 406—409.

Nabi, H., Koskenvuo, M., Singh-Manoux, A., Korkeila, J., Suominen,S., Korkeila, K., Vahtera, J., Kivimaki, M., 2010. Low pessimismprotects against stroke: the Health and Social Support (HeSSup)prospective cohort study. Stroke 41, 187—190.

Needham, B.L., Adler, N., Gregorich, S., Rehkopf, D., Lin, J., Black-burn, E.H., Epel, E.S., 2013. Socioeconomic status, health beha-vior, and leukocyte telomere length in the National Health andNutrition Examination Survey, 1999—2002. Soc. Sci. Med. 85, 1—8.

Nilsson, P.M., 2011. Telomeres and cardiovascular disease: newperspectives in hypertension. J. Hum. Hypertens. 25, 703—704.

Nordfjall, K., Eliasson, M., Stegmayr, B., Melander, O., Nilsson, P.,Roos, G., 2008. Telomere length is associated with obesity para-meters but with a gender difference. Obesity 16, 2682—2689.

Novotny, P., Colligan, R.C., Szydlo, D.W., Clark, M.M., Rausch, S.,Wampfler, J., Sloan, J.A., Yang, P., 2010. A pessimistic explana-tory style is prognostic for poor lung cancer survival. J. Thorac.Oncol. 5, 326—332.

O’Donovan, A., Lin, J., Dhabhar, F.S., Wolkowitz, O., Tillie, J.M.,Blackburn, E., Epel, E., 2009. Pessimism correlates with leuko-cyte telomere shortness and elevated interleukin-6 in post-meno-pausal women. Brain Behav. Immun. 23, 446—449.

Peters, J.L., Kubzansky, L.D., Ikeda, A., Spiro 3rd., A., Wright, R.O.,Weisskopf, M.G., Kim, D., Sparrow, D., Nie, L.H., Hu, H.,Schwartz, J., 2011. Childhood and adult socioeconomic position,cumulative lead levels, and pessimism in later life: the VA Nor-mative Aging Study. Am. J. Epidemiol. 174, 1345—1353.

Roy, B., Diez-Roux, A.V., Seeman, T., Ranjit, N., Shea, S., Cushman,M., 2010. Association of optimism and pessimism with inflamma-tion and hemostasis in the Multi-Ethnic Study of Atherosclerosis(MESA). Psychosom. Med. 72, 134—140.

Sampson, M.J., Winterbone, M.S., Hughes, J.C., Dozio, N., Hughes,D.A., 2006. Monocyte telomere shortening and oxidative DNAdamage in type 2 diabetes. Diabetes Care 29, 283—289.

Scheier, M.F., Carver, C.S., 1985. Optimism, coping, and health:assessment and implications of generalized outcome expectan-cies. Health Psychol. 4, 219—247.

Scheier, M.F., Carver, C.S., Bridges, M.W., 1994. Distinguishing opti-mism from neuroticism (and trait anxiety, self-mastery, and self-esteem): a reevaluation of the Life Orientation Test. J. Pers. Soc.Psychol. 67, 1063—1078.

Schou, I., Ekeberg, O., Sandvik, L., Ruland, C.M., 2005. Stability inoptimism—pessimism in relation to bad news: a study of womenwith breast cancer. J. Pers. Assess. 84, 148—154.

Slagboom, P.E., Droog, S., Boomsma, D.I., 1994. Genetic determina-tion of telomere size in humans: a twin study of three age groups.Am. J. Hum. Genet. 55, 876—882.

von Zglinicki, T., 2002. Oxidative stress shortens telomeres. TrendsBiochem. Sci. 27, 339—344.

Weng, N.P., Levin, B.L., June, C.H., Hodes, R.J., 1996. Regulatedexpression of telomerase activity in human T lymphocyte devel-opment and activation. J. Exp. Med. 183, 2471—2479.

Wong, J.M., Collins, K., 2001. Telomere maintenance and disease.Lancet 362, 983—988.