Fifteen-Year Follow-up of Association Between Telomere Length and Incident Cancer and Cancer Mortality FREE

To the Editor: Telomeres are nucleoprotein structures that cap the ends of chromosomes and confer chromosomal stability. Telomere shortening to a critical length due to extensive DNA replication or oxidative stress facilitates genomic mutations and may induce malignant transformation.¹

We have previously shown that low leukocyte telomere length (TL) is associated with an increased risk of cancer incidence and mortality.² Although the associations were of considerable strength, independent of standard risk factors, and consistent in subgroups, the number of incident cases of cancer was less than 100. We extended the follow-up from 10 to 15 years, increased the number of events to 137, and added data from a second TL measurement because a single measurement may underestimate the true association.

A prospective population-based cohort study was conducted on a random sample of 1000 persons aged 40 to 79 years in Bruneck, Italy, beginning in 1990 with clinical and laboratory evaluations every 5 years. The study was approved by the ethics committee of Bolzano province, and all participants provided written informed consent. This evaluation focused on the period between the first reexamination and fourth reexamination (1995-2010). Telomere length was available for 787 participants free of cancer in 1995 (96.9%) and 558 participants in 2005 (98.1%). Telomere length was assessed in leukocytes from fasting blood samples with quantitative polymerase chain reaction, and values were corrected for reaction efficiency.² Incident cancer and cancer-related death as of October 2010 were ascertained by review of medical records, including histopathological workup and death certificates containing detailed information on cause and circumstances of death (follow-up rate, 100%).²

To quantify within-person variability of TL and C-reactive protein level, regression dilution ratios (RDRs) were calculated based on 1995 (baseline) and 2005 values.³ Long-term average (“usual”) TL based on TL measured in 1995 and 2005 was estimated by multivariate regression calibration.³ Separate Cox models using baseline and usual TL to predict cancer incidence and cancer mortality over the 15-year period were fitted for TL tertile groups and continuous log_e TL. The proportional hazard assumption was confirmed using Schoenfeld residuals. A 2-sided P < .05 was considered significant. Statistical analyses were performed with Stata version 10.1 (StataCorp, College Station, Texas).

Distribution of baseline variables and their correlations with TL were outlined in the previous report.² The age- and sex-adjusted RDR of TL was 0.59 (95% confidence interval [CI], 0.52-0.65) comparable with that of C-reactive protein (RDR, 0.56; 95% CI, 0.48-0.64).

During follow-up, 137 of 787 participants were diagnosed with cancer (incidence rate, 14.3 per 1000 person-years; 95% CI, 12.1-16.9) and 62 of 787 participants died from cancer. Incidence rates of cancer in the longest, middle, and shortest TL tertiles were 5.9 (95% CI, 3.9-9.0), 16.9 (95% CI, 12.9-22.2), and 22.8 (95% CI, 17.8-29.2), respectively. The adjusted hazard ratio per 1-SD lower baseline TL (SD = 0.52) was 1.56 (95% CI, 1.32-1.85) and 1.88 (95% CI, 1.48-2.40) for cancer incidence and cancer mortality, respectively (Table). Cox models including usual instead of baseline TL yielded hazard ratios of 2.15 (95% CI, 1.61-2.88) and 2.98 (95% CI, 1.96-4.53) for cancer incidence and cancer mortality, respectively. Cumulative hazard plots by tertiles of baseline TL are shown in the Figure.

This 15-year follow-up corroborates our previous findings² that short telomeres are associated with cancer incidence and cancer mortality. An RDR of 0.59 and the stronger associations for usual TL underscore the importance of telomere dynamics in carcinogenesis and the need for multiple measurements of TL in the characterization of individual cancer risk. The variability of TL over time is similar to previous evaluations on telomere dynamics^4- 5 and may reflect the cumulative effect of environmental and behavioral exposures, varying telomerase activity, and stress-induced repopulation of peripheral blood by recently dividing hematopoietic bone marrow cells.⁶ Additional variability may arise from shifts in the cellular composition of peripheral blood leukocytes. Limitations include that the population was entirely white, TL measurements were available in leukocytes only, and power to analyze individual cancer types was limited.