Race Differences in Telomere Length in Benign Prostate Biopsies and Subsequent Risk of Prostate Cancer

Prostate carcinogenesis is a multistep evolutionary process that includes the acquisition of initiating genetic and epigenetic alterations that confer a proliferative advantage to preneoplastic cells. Increasing evidence supports that alterations in the surrounding tissue microenvironment promote prostate carcinogenesis (1). Therefore, interrogation of precancerous cellular changes that may predispose to a malignant phenotype has value in terms of primary prevention and greater insight into the full spectrum of carcinogenesis.

Tissue-based measurements of telomeres, the repetitive DNA elements located at chromosomal ends that are essential for maintenance of genomic integrity, is a potential biomarker of early malignancy (2, 3). Evidence exists for both shortened and lengthened telomeres having a role in cancer. The shortening of telomeres is known be associated with aging (4) and genetic instability, a harbinger of cancer (1). Telomere shortening is likely a result of oxidative stress (5), a byproduct of a putative cancer risk factor, chronic inflammation (6, 7), and has been proposed as a field cancerization marker in the prostate (3). However, telomere shortening also is considered a protective mechanism against tumorigenesis; cells with longer telomeres have greater proliferative potential and likelihood of accruing mutations (3). In addition, the ability of a cell to maintain telomere length, either through reactivation of telomerase or via the alternative lengthening of telomeres pathway, is also a hallmark of cancer (3).

Several epidemiologic studies suggest that longer telomeres may increase risk for cancer based on associations between genetic variation that influences telomere length and cancer risk (8–11). However, a genetic predisposition to longer telomere length is not consistently associated with increased cancer risk and the few studies that have specifically studied prostate cancer in this regard have equivocal results (8, 12, 13). In addition, other investigations have suggested that short telomeres associate with poor patient survival in different cancers (14–18). A meta-analysis suggested that shorter telomeres in prostate had a positive association with disease risk, but in leucocytes shorter telomere length had a negative association (19). A more recent study not included in this meta-analysis found leukocyte telomere length was shorter in prostate patients with high Gleason scores and that both measured and genetically predicted short leukocyte telomere length are associated with worse prognosis in patients with prostate cancer receiving definitive treatment (20).

Telomere length has also become an active area of investigation in terms of a potential biologic explanation for cancer health disparities (21–23). African Americans (AA) are known to have longer leukocyte telomere length than Whites at birth (24), but may also have increased telomere length attrition over time (25) with older AAs displaying shorter telomere lengths than their Whites counterparts (26). Compared with White men, AA men have a higher age-adjusted prostate cancer incidence and mortality rates and have worse outcomes following primary treatment of their disease (27). The known prostate cancer precursor, high-grade prostatic intraepithelial neoplasia (HGPIN), is more prevalent in AA men (28, 29) and may increase risk for prostate cancer even in AA men with low PSA (30). Evidence exists for high levels of telomere dysfunction in HGPIN similar to what is observed in prostate tumors (31). In addition, our group has previously shown that polycyclic aromatic hydrocarbon–DNA adducts, a biomarker of DNA damage (32), has a disproportionate effect on prostate cancer risk for AA men when found at higher levels in benign prostate (33). Interestingly, a recent study found an association between higher total PSA levels and longer leukocyte telomere length that was strongest in non-Hispanic blacks (34).

To date, only one study comparing prostate tissue–based telomere length by race has been performed and found no differences by race but did find that in men with high-grade disease a greater proportion of AA men had an adverse prostate cancer cell telomere phenotype (22). In addition, there is only one report of telomere length in premalignant prostate tissue and subsequent prostate cancer risk, albeit in prostate biopsies with HGPIN. To evaluate the role of prostate telomere length for risk of subsequent prostate cancer and whether race differences exist in telomere length and prostate cancer risk, we studied a large ethnically diverse matched case–control sample nested within a retrospective cohort of men with a benign biopsy. Our study population allowed us to examine race differences in premalignant prostate telomere length as well as whether premalignant prostate telomere length associated with risk of developing prostate cancer.

The study was approved by the Henry Ford Health System Institutional Review Board (IRB) and conducted in accordance with ethical guidelines outlined in the Belmont Report. Because data collection only involved a medical record review, informed consent was waived by the IRB. A nested case–control study sample was drawn from a historical cohort of 10,478 men with a benign prostate needle core biopsy surgical specimen between January 1990 and December 2012. Details of study eligibility criteria and case–control matching are described previously (35). A total of 672 of the 725 eligible cases constituted matched case–control pairs with prostate biopsy tissue available for assaying. The remaining 53 cases could not be matched to form an eligible case–control pair due to (i) cancer detected in pathologic reexamination of recut “benign” biopsy at the time of the cohort entry (n = 19, 36.5%), (ii) prior history of prostate (n = 5, 9.6%) or other types of cancer (n = 4, 7.7%) found upon medical record review, (iii) lack of eligible controls that could be matched to a case (n = 15, 28.3%), (iv) no analyzable prostate tissue (n = 8, 15.4%), and (v) prostate cancer diagnosis could not be confirmed (n = 2, 3.8%). Within these eligible 672 case–control pairs, telomere assays were run on the first 597 eligible matched pairs with analyzable telomere length results available for 920 unique individuals and complete telomere and covariate data available for 524 case–control pairs (Supplementary Fig. S1). All benign tissue specimens were evaluated for the presence of cancer, HGPIN, atrophy, and inflammation by a genitourinary pathologist (N.S. Gupta or S.R. Williamson) blinded to case–control status as described previously (35).

Biopsy specimens preserved as formalin-fixed paraffin-embedded (FFPE) blocks were procured from the Henry Ford biorepository and serial sections at 5 µmol/L thickness were cut. The middle section was stained with hemoxylin and eosin to confirm the tumor/benign status of the specimen and for histopathologic assessment of atrophy and inflammation. Two unstained curls from each specimen were collected for the telomere length assay. Approximately 25 ng of DNA was extracted from FFPE prostate biopsy tissue curls using the QIAamp DNA FFPE tissue protocol (Qiagen). Briefly, this involved a six-step process where the tissue-embedded paraffin was dissolved in xylene and removed, lysed under denaturing conditions with proteinase K, heated at 90°C to reverse formalin cross-linking and then purified by binding to a membrane to remove contaminants that are washed away and concentrated DNA is eluted from the membrane. Telomere length was assessed using a modified qRT-PCR that correlates with Southern blot measures of telomere length (36, 37). Two reactions were prepared, one for telomere amplification and one for the reference albumin gene (primers listed in Supplementary Table S1). Telomere (T) and single-copy gene (S) PCRs were performed using 7500 Real-Time PCR System with 96-well plates in separate 96-well plates (Applied Biosystems). To minimize batch effects, matched cases and controls were assayed on the same T and S plate pairs. Quantification of unknowns was determined by comparison with a standard curve derived from amplification, in the same plate, of serial dilutions of a reference DNA sample derived from FFPE prostate tissue from a prostatectomy specimen. The average PCR efficiency (E) and R² for the standard curves were: telomere E = 100.7%, R² = 0.964 and albumin: E = 87.4%, R² = 0.967. The thermal cycling profile for both telomere and single-copy gene started with 95 °C incubation for 3 minutes, followed by 40 cycles of 3-second denaturation at 95 °C and 20-second annealing/extension at 60 °C. Three identical aliquots of each DNA sample were processed in parallel on the T and S plates with an averaged final result.

Telomere length measures were subject to a cube root transformation to make the overall data distribution more closely conform to normality. To take assay variation into account, all telomere analyses were weighted by the inverse coefficient of variation based on the three assays run for each specimen. Analyses of the association of telomere length with race and other potential confounders were done cross-sectionally using all available telomere data (n = 920 individuals). Because this dataset was derived from a matched case–control sampling scheme, an inverse probability weighting scheme was used to account for sampling bias. Among all potential controls, a logistic regression was used to estimate for the probability of a control subject being selected as part of the analytic dataset based on matching variables. For cases, the probability was derived as the proportion of cases represented in the analytic dataset (n = 571) over the total number of cases in the cohort (n = 1,714). Case–control clusters were also identified on the basis of the matching scheme to account for the correlation structure across matched sets. Cross-sectional analyses were then conducted using a general estimating equations (GEE) model with observations weighted by the inverse sampling probabilities described above multiplied by the inverse coefficient of variation of the telomere assay with an exchangeable correlation structure across clusters of cases and controls.

Initial case–control comparisons of telomere length used paired t tests. Modeling of associations with prostate cancer case status was done using conditional logistic regression analyses that accounted for the one-to-one matched case–control design—case–control pairs were matched on age (±1 year), race, and date of cohort entry (±1 year). Associations with telomere length were examined both on a continuous and categorical scale—the latter as quartile groups. Quartile groups were established on the basis of the telomere length distribution in controls and cut-off points were made in a race-specific manner. Initial univariable models assessed associations between prostate cancer case status and telomere length, overall and stratified by race. Next, multivariable models were fitted to jointly model telomere length while including clinical variables. Finally, a series of multivariable models were fit stratified on race and case aggressive status.

The data generated in this study are available upon request from the corresponding author.

Table 1 summarizes the analytic study sample which was comprised of 53.1% White case–control pairs and 46.9% of AA case–control pairs. In addition to race, case–control pairs were matched on age and date of cohort entry. Cases had significantly higher PSA levels at cohort entry (6.8 ± 6.0 ng/mL vs. 5.6 ± 3.5 ng/mL; P < 0.0001) and number of PSA tests between the time of cohort entry and diagnosis (4.9 ± 4.9 tests vs. 4.3 ± 4.1 tests; P = 0.02). While cases had more HGPIN (P = 0.05), they had less benign prostatic hypertrophy (BPH; P = 0.002), clinical prostatitis (P = 0.02), and simple atrophy (P = 0.05) than controls. Cases also had less prostatic inflammation (60.7% vs. 63.8%), but this difference was not statistically significant. Mean telomere length was not different between cases and controls (1.46 ± 0.38 vs. 1.45 ± 0.42; P = 0.49).

Telomere length is known to be different by race (21, 26), and this held true in our prostate telomere length data with White men having significantly longer telomeres in benign prostate compared with AA men (Fig. 1). After adjusting for batch effects and weighting by the inverse sampling probability and coefficient of variation for the telomere assay the differences in mean telomere length by race were even more profound (1.51 ± 0.38 for AA vs. 1.63 ± 0.39 for Whites; P < 0.0001). Using the data from our unmatched sample, that was supplemented by telomere length data of cases and controls not used in the matched analysis because the pair had incomplete telomere data (Supplementary Fig. S1), we further explored the relationship of telomere length with clinical and demographic variables in the context of race (Table 2). On a univariate level, in addition to race, age (P = 0.04) and simple atrophy (P = 0.002), were significantly associated with telomere length with age associated with decreasing telomere length and simple atrophy associated with increased telomere length. Adjusting for race reduced both associations and neither were significantly associated after the race adjustment. Adjusting for race did make some associations with telomere length stronger, notably BPH and HGPIN, although neither reached statistical significance. For HGPIN, a negative association with telomere length was only observed in White men (β = 0.859; P = 0.08). Interestingly, ever smoking had a nominal negative association with telomere length (β = 0.899; P = 0.16) that was reduced after adjusting for race. Further investigation of the association between ever smoking and telomere length in a race-specific manner found ever smoking AA men had a significantly lower telomere length than nonsmokers (P = 0.04). Of all the factors examined in Table 2, only PSA levels were significantly different by race (Supplementary Table S2).

Considering the observed race differences in prostatic telomere length and positing that if telomere length has an effect on prostate cancer risk, it may not be linear, telomere length quartiles and prostate cancer risk in the full and race-stratified samples were analyzed (Table 3). In an adjusted model, risk of prostate cancer nominally increased with each successive quartile of telomere length compared with the lowest quartile (P = 0.03). Men in the highest quartile of telomere length had the highest prostate cancer risk [OR = 1.79; 95% confidence interval (CI) = 1.16–2.74]. In adjusted models, White men had successively greater risk of prostate cancer with increasing telomere length—telomere length in the highest quartile was associated with a 90% increased risk of prostate cancer (OR = 1.90; 95% CI = 1.03–3.36). While prostate cancer risk was elevated in the three higher telomere length quartiles relevant to the lowest quartile in AA men, less evidence existed for a trend of increasing prostate cancer risk being associated with increasing telomere length.

We next investigated whether prostate cancer risk associated with telomere length differed upon stratification for aggressive disease (Table 4). Stratifying cases on tumor grade and a more inclusive definition of disease aggressiveness that was defined as Gleason grade group 3 or higher or stage 3A or higher or PSA level at diagnosis of 20 ng/mL or above, prostate cancer risk associated with telomere length was greater in cases classified as more aggressive by either definition. Men with telomere lengths in the fourth quartile had a two and a half-fold greater risk for high-grade prostate cancer (OR = 2.63; 95% CI = 1.07–6.51). Using the more inclusive definition of aggressive disease that included tumor stage and PSA level showed similar results, with ORs and significance levels nominally higher compared with aggressive cases defined only by tumor grade.

Finally, we reexamined the associations of telomere length with aggressive prostate cancer in a race-stratified manner (Table 5). Associations for both definitions of aggressive disease were of slightly greater magnitude in White men than observed in the whole study population. For instance, White men with telomere lengths in the fourth quartile had an OR of almost three for aggressive prostate cancer (OR = 2.98; 95% CI = 1.03–8.56) as well as a suggestive trend for increasing telomere length being associated with risk for aggressive prostate cancer (OR_trend = 1.37; 95% CI = 1.00–1.88). For high-grade disease, similar associations were observed in White men. The risk for aggressive and high-grade prostate cancer for AA men was also increased for prostate telomere lengths in the second to fourth quartiles compared with the referent first quartile, but less evidence existed for a trend between increasing risk and telomere length.

While numerous studies have previously investigated telomere length in prostate cancer, most have utilized the surrogate measure of leukocyte telomere length (20, 34, 38–43) whereas the investigation of telomere length in prostate cells has largely been limited to prostates of men with cancer (2, 5, 22, 31, 44–51). We used a matched case–control sample nested in a cohort of men with a benign prostate biopsy to investigate the relationship between prostate cancer susceptibility and telomere length in prostate before onset of disease. While no difference in average telomere length in benign prostate biopsies between cases and controls was detected, nonlinear models based on quartile groupings of telomere length showed evidence for an incremental higher prostate cancer risk for men with increasing telomere length quartile.

The rapid cell division characteristic of cancer would naturally result in telomere shortening. However, most tumors avoid telomere shorting and the accompanying cellular senescence by activating telomerase, a ribonucleoprotein with reverse transcription activity that adds de novo telomere hexanucleotide repeats to the chromosomal ends and counteracts telomere erosion thereby preventing cells from entering senescence (52, 53) and is activated in prostate cancer (54, 55). Increased telomerase activity has been detected in the prostate cancer precursor, HGPIN (54), and when present may increase risk of prostate cancer development (56). Two studies have also measured HGPIN and telomere length in the prostate (57, 58). Comparators were different in the two studies, but consistent with our suggestive findings in White men both found a reduced telomere length associated with HGPIN—with one study comparing telomere length in HGPIN to benign prostate epithelial cells (57) and the other to the surrounding stroma. In the Joshua and colleagues study, a cohort of 68 men without prostate cancer but who had HGPIN on prostatic biopsy were followed forward for development of cancer. Contrary to our results, they found reduced telomere length in cells with HGPIN was predictive for development of cancer. In our study sample, less than 10% of men had HGPIN, whereas HGPIN was present in 100% of the sample in aforementioned study (58). In addition, telomere length was measured directly in cells with HGPIN and surrounding stroma by fluorescence (58) compared with the nonspecific qPCR method used in the current study.

Several studies have tested the association between underlying genetic variation influencing telomere length and prostate cancer risk and found suggestive associations between genetic variation associated with longer telomeres and increased prostate cancer risk (8, 9). However, the reported increased risk has been small (10%–20%) and generally not statistically significant. In addition, these genetic studies did not perform race-specific analyses, so it is unclear whether this nominal risk associated with a genetic predisposition to longer telomeres was consistent across race groups. In our study, increasing telomere length had a stronger relationship with aggressive prostate cancer risk. In tumor cells, Heaphy and colleagues (59) found the proportion of cases with normal/long telomeres significantly increased with increasing tumor grade group and pathologic stage but in an earlier report found shorter telomere length in prostate cancer–associated stromal cells was associated with a greater likelihood to progress to metastasis or die of prostate cancer (46). Longer telomeres in leukocytes have been more often reported associated with unfavorable (40–43) rather than favorable (20, 44) prostate cancer outcomes.

Telomeres in benign prostate biopsies were significantly shorter in AA compared with White men. While race differences in leukocyte telomere length have been widely reported (21, 24, 26, 60, 61), only one other study has examined race differences in telomere length in relation to prostate cancer risk. Heaphy and colleagues reported a greater proportion of Black men with higher grade disease had an adverse prostate cancer cell telomere phenotype (as defined by variable telomere length) than White men with high-grade disease (22) but reported no difference in the prevalence of short telomeres in prostate stromal cells by race. Another recent study showed that leukocyte telomere length was shorter in AA patients with prostate cancer with higher Gleason grade and also associated with worse prognosis, but this study was limited to AAs and hence could not do any race comparisons. We also found that cigarette smoking was associated with decreased telomere length in benign prostate, but only in AA men. This association is consistent with previous reports of exposure to tobacco being associated with shorter telomeres (62, 63) including one report where cigarette smoke exposure appears to have a greater adverse effect on telomere length in Black women (61). Another study of White men found cigarette smoking history was not associated with telomere length in prostate cancer or benign stromal cells (64). In our study, history of cigarette smoking was not associated with prostate cancer risk. The question of racial differences with regard to telomere length and prostate cancer risk warrants further examination especially if the telomere complex becomes a therapeutic target in prostate cancer (65).

Given our experimental approach that involved measuring telomeres in bulk DNA from prostate biopsies, it is not possible to do direct comparisons of our study results with prostate cancer studies that measured telomere length based on surrogate leukocyte DNA, genetic inheritance, or single-cell FISH methods. Furthermore, because all cohort members had at least one prostate biopsy they were at elevated risk for prostate cancer and are likely unrepresentative of all men in the same age and race demographic. Men diagnosed with cancer a year or less after cohort entry were excluded to minimize the number of cohort members with undetected prostate cancer. Nonetheless, based on the age and risk profile of men in our cohort some likely had synchronous prostate cancer that was missed on biopsy. Our nested case–control sample study design allowed us to estimate the prostate cancer risk associated with telomere length in prostate before onset of disease—only a few studies have measured telomeres in prostate tissue and all but one of these have been after cancer diagnosis (22, 44–46, 50, 51, 58). In using the entire tissue specimen rather than separate cell types, telomeres from both prostate epithelial and stromal cells as well as any immune cells that were present in the tissue specimen were measured. In a nonhyperplastic prostate, epithelial and stromal cells are approximately found in equal proportion (66); therefore, we are unable to infer whether telomere length in prostate stromal or epithelial cells (or both) may influence prostate cancer risk. However, because there was no association between prostate inflammation and telomere length, it is unlikely that differential immune cell populations had an influence on telomere length results. Finally, by accounting for batch effects and matching on specimen age, we in part addressed issues related to sample type selection, sample collection, storage, processing issues, and assay procedures that can influence results (67, 68).

In summary, we report for the first time on a study of telomere length in premalignant prostate tissue and subsequent prostate cancer risk. In White men, there was a significant increased risk trend associated with longer telomeres and this association was enhanced when analyses were restricted to aggressive cases which add further credence to these results. In addition, White men were found to have significantly longer telomeres in benign prostate biopsies compared with AA men. Studies using alternative methods to measure telomeres in premalignant tissues, principally those that can assess telomere length on a cell-to-cell basis (48), may shed some more light on role telomeres play in early prostate carcinogenesis.

B.A. Rybicki reports grants from NIH during the conduct of the study; in addition, B.A. Rybicki is an expert epidemiology consultant for Downs Law Group in a lawsuit against BP Oil for the Deepwater Horizon Oil spill. No disclosures were reported by the other authors.

B.A. Rybicki: Conceptualization, resources, supervision, validation, methodology, writing–original draft. S.M. Sadasivan: Resources, data curation, methodology, project administration, writing–review and editing. Y. Chen: Formal analysis, validation, investigation, methodology, writing–review and editing. I. Loveless: Formal analysis, validation, investigation, methodology, writing–review and editing. N.S. Gupta: Investigation, visualization, methodology, writing–review and editing. D.A. Chitale: Investigation, visualization, methodology, writing–review and editing. S.R. Williamson: Investigation, visualization, writing–review and editing. A.G. Rundle: Investigation, visualization, methodology, writing–review and editing. D.L. Tang: Conceptualization, resources, supervision, validation, investigation, visualization, methodology, writing–review and editing.

This work is supported by NIH 5R01-ES011126 awarded to B.A. Rybicki.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.