Leukocyte telomere length in relation to risk of lung adenocarcinoma incidence: Findings from the Singapore Chinese Health Study

Abstract

Telomeres are crucial in the maintenance of chromosome integrity and genomic stability. Critically short telomeres can trigger programmed cell death while cells with longer telomeres may have increased likelihood of replicative errors, resulting in genetic mutations and chromosomal alterations, and ultimately promoting oncogenesis. Data on telomere length and lung cancer risk from large prospective cohort studies are spare. Relative telomere length in peripheral blood leukocytes was quantified using a validated monochrome multiplex quantitative polymerase chain reaction (qPCR) method in 26,540 participants of the Singapore Chinese Health Study. After a follow-up of 12 years, 654 participants developed lung cancer including 288 adenocarcinoma, 113 squamous cell carcinoma, and 253 other/unknown histological type. The Cox proportional hazard regression was used to estimate hazard ratio (HR) and 95% confidence interval (CI). HR of lung adenocarcinoma for individuals in the highest comparing the lowest 20 percentile of telomere length was 2.84 (95% CI 1.94–4.14, P_trend<0.0001). This positive association was present in never smokers (P_trend<0.0001), ever smokers (P_trend=0.0010), men (P_trend=0.0003), women (P_trend<0.0001), and in shorter (P_trend=0.0002) and longer (P_trend=0.0001) duration of follow-up. There was no association between telomere length and risk of squamous cell carcinoma or other histological type of lung cancer in all or subgroups of individuals. The agreement of results from this prospective cohort study with those of previous prospective studies and Mendelian randomization studies suggest a possible etiological role of telomere length in the development of lung adenocarcinoma.

INTRODUCTION

Lung cancer is the leading cause of cancer death in both men and women in the U.S. and worldwide.^{1, 2} It is estimated that more than 155,000 Americans will die from lung cancer in 2017.³ Although the overall 5-year survival rate of lung cancer patients is only 18%, patients with localized lung cancer at diagnosis have a 5-year survival rate of 55%.³ Since 1970s, the number of lung adenocarcinomas has increased from approximately 20% to around 50%, and has surpassed squamous cell carcinoma, becoming the most common histological subtype of lung cancer in early 2010s.⁴ With the increasing occurrence of lung cancer adenocarcinoma cases, identifying novel markers for lung adenocarcinoma is important for understanding the increasing incidence of this malignancy.

Telomeres consist of long tracts of nucleotide repeats (TTAGGG) and are associated with a protein complex termed shelterin at the ends of chromosomes that is essential for maintaining chromosomal integrity.^{5, 6} Telomeres shorten with each cell division due to the end replication problem.⁷ In stem and progenitor cells, the telomerase enzyme elongates telomeres, enabling prolonged cell survival and proliferation.⁸ Telomerase is also activated in more than 90% of human cancers,⁹ resulting in sustained proliferation and survival of cells that otherwise would have undergone irreversible growth arrest or programmed cell death.¹⁰ In this way, telomere shortening in telomerase deficient normal cells serves as a mechanism of tumor suppression.

Recently, Mendelian randomization studies using germline genetic variants as instrumental variables derived from genome-wide association studies showed that a genetic score for longer telomere length was significantly associated with increased risk of lung cancer.^11–13 These findings are intriguing since they imply causal relevance. Given the potential limitations due to pleiotropy effect, population stratification, or ancestry from the Mendelian randomization studies, we examined the association between directly quantified telomere length and the risk of lung cancer incidence in the Singapore Chinese Health Study. Our findings of a statistically significant, robust association between longer telomeres and increased risk of lung adenocarcinoma, along with previous prospective and Mendelian randomization studies, strongly support to a possible etiological role of telomere length in the development of lung adenocarcinoma.

MATERIALS AND METHODS

Study Population

Study subjects were drawn from the Singapore Chinese Health Study.¹⁴ Briefly, from April 1993 through December 1998, the Singapore Chinese Health Study enrolled 63,257 Chinese men and women aged 45–74 years who belonged to the Hokkien or Cantonese dialect group, and resided in government-built housing estates. At recruitment, each subject was interviewed in person by a trained interviewer to obtain information on demographics, body weight and height, lifetime use of tobacco (cigarettes and water-pipe), current physical activity, menstrual/reproductive history (women only), occupational exposure, medical history, and family history of cancer. Information on current diet and consumption of beverages was assessed via a 165-item food frequency questionnaire that had been validated against a series of 24-hour dietary recall interviews and selected biomarker studies conducted on random subsets of cohort participants.^15–17 Blood and urine samples were initially collected from a 3% random sample of cohort participants in 1994–1999. Beginning in 2000, we extended blood and urine collection to all surviving cohort participants. Of the 52,322 eligible subjects, 28,346 subjects (54% of eligible) donated blood samples. The study participants who provided blood samples were younger (mean ± standard deviation: 60.9 ± 7.7 versus 62.4 ± 8.2 years), more educated (33.6% versus 25.1% having secondary or higher education), and more likely to be men than those who did not provide biospecimens (45.5% versus 39.1%), and were otherwise similar in smoking (32% versus 30.2% ever smokers) and alcohol consumption (18.3% versus 14.8% weekly consumers of alcohol). The present study was approved by the Institutional Review Boards of the respective institutions.

Assessment of Lung Cancer Cases

Identification of incident cases of cancer and death was accomplished by annual record linkage of all surviving cohort participants with the database of the nationwide Singapore Cancer Registry and the Birth and Death Registry that have complete records of incident cancer and death cases, respectivley.¹⁸ To date, only 47 (<1%) of the entire cohort participants were known to be lost to follow-up due to migration out of Singapore, suggesting that the ascertainment of cancer and death incidences among the cohort participants was virtually complete. As of December 2015, with an average follow-up of 11.8 years after their donation of blood sample, 654 developed lung cancer among 26,761 subjects after excluding 1,585 participants with a history of cancer at the time of blood collection. Of the 654 lung cancer cases, 573 (87.6%) were histopathologically confirmed while the remaining 81 (12.4%) were diagnosed with radiography or computer-assisted tomography evidence. Among the histopathologically confirmed cases, 279 (48.7%) were adenocarcinomas, 113 (19.7%) were squamous cell carcinomas, 56 (9.8%) were small cell cancers, and 125 (21.8%) were other cell types.

Measurements of Telomere Length

Genomic DNA was extracted from peripheral blood using QIAamp 96 DNA Blood kits (Qiagen, Valencia, CA) according to the manufacturer’s protocol. Telomere length was measured using a validated monochrome multiplex qPCR method.¹⁹ This method measures the relative average telomere length in genomic DNA by determining the ratio of telomere repeat copy number (T) to single (albumin) gene copy number (S) in experimental samples relative to a reference sample. The DNA sample for the standard curve was composed of an equimolar pool of 77 samples selected from participants of the Singapore Chinese Health Study who were identified in a prior study; the telomere length values of all the 77 samples were within 10% of the population mean. This pooled DNA sample was run on all qPCR plates: 8 replicates for each of four concentrations (4, 0.8, 0.16, and 0.032 ng/μl). Thermal cycling was carried out on an Applied Biosystem 7900 HT instrument, using PCR cycling conditions as described.¹⁹ Real-time PCR cycle thresholds, determined independently for the albumin gene (ALB) and telomere (TEL) amplification traces for all wells (experimental and standard DNA samples), were used to calculate telomere length as described previously¹⁹ with the 384-well plate-based normalization of telomere length, which was more robust than the overall standard-curve based normalization. All experimental DNA samples were assayed in duplicate, and the average value of the two replicates was used for final analysis for each subject. The mean coefficient of variation, as a measure of reproducibility, of all technical sample duplicates for telomere length in the present study was 3.5%.

Statistical Analysis

The χ² test was used to compare the distributions of selected variables between lung cancer case groups and non-cases whereas The t test and analysis of covariance (ANCOVA) method were used to examine the difference in telomere lengths between two groups or among 3 or more groups.

The present analysis included 26,540 subjects after excluding 221 subjects with unavailable telomere length measurement due to assay problems. For each subject, person-years at risk were computed from the date of blood draw to the date of lung cancer diagnosis, death, migration out of Singapore, or December 31, 2015, whichever occurred first. Cox proportional hazard regression method was employed for calculation of hazard ratios (HRs) and their corresponding 95% confidence intervals (CIs) of lung adenocarcinoma, squamous cell carcinoma, and other or unknown histological types of lung cancer separately associated with higher telomere length quintiles comparing with the lowest quintile of telomere length. Test for linear trend was conducted by treating the quintiles of telomere length as an ordinal variable in the Cox model. The proportional hazards assumption was examined and valid. In all analyses, we adjusted for smoking history by including categorical terms of average number of cigarettes smoked per day (never smokers, 1–12, 13–22, or 23+), number of years of smoking (never smokers, 1–19, 20–39, or 40+), and number of years since last smoked for quitters (current smokers, <1, 1–4, 5–19, 20+ years since last smoked, or never smokers). Other potential confounders included in the multivariate Cox proportional hazards models were age, sex, dialect group (Hokkiens or Cantonese), level of education (no formal education, primary school, secondary or higher education), body mass index (<20, 20-<24, 24-<28, or ≥28 kg/m²), and alcohol consumption (Non-drinkers, <7, or ≥7 drinks per week). Sensitivity analyses were performed for never smokers versus ever smokes, men versus women, and for longer versus shorter duration of follow-up.

The cumulative incidence rates of lung adenocarcinoma for individuals with shorter and longer telomeres were estimated using a competing risk method described by Gaynor et al.²⁰ We estimated the cumulative incidence rate of lung adenocarcinoma for smokers and non-smokers separately with short (the lowest quintile of telomere length), intermediate (the 2^nd to 4^th quintiles of telomere length) and long telomeres (the highest quintile of telomere length) adjusted for competing risk of death using a SAS Software Macro described by Beiser et al.²¹

Statistical analyses were conducted using SAS software version 9.4 (SAS Institute, Cary, NC). All P-values reported are two-sided, and those less than 0.05 were considered statistically significant.

RESULTS

The mean age (standard deviation) at cancer diagnosis of all lung cancer patients (n = 654) was 73.2 (7.0) years. The median time interval between blood draw and cancer diagnosis was 6.8 years (range: 1 month to 21.1 years).

The distributions of selected demographics and lifestyle factors in lung cancer patients compared with the remaining participants of the cohort are shown in Table 1. Compared with non-cases, lung cancer cases overall had significantly higher percentages of former (21.1%) and current (48.5%) smokers (P < 0.001). Among ever smokers, lung cancer patients reported more numbers of cigarettes smoked per day, years of smoking, and pack-years of smoking than non-cases (all P values < 0.001). The percentage of regular alcohol drinkers was slightly higher in lung cancer cases than in non-cases (Table 1). Among 3 histological groups of lung cancer, patients with adenocarcinoma were younger, more likely to be women and never smokers compared to those with squamous cell carcinoma or other/unknown histological types of lung cancer (Table 1).

Table 1.

Characteristics of lung cancer cases by common histology type comparing with non-cancer cases The Singapore Chinese Health Study 1993–2015

Characteristics*	Non-cancer	Adenocarcinoma	Squamous cell cancer	Cancer of other or unknown histology	P†	P‡
Number of subjects	25,886	288	113	253	…	…
Mean age (SD), years	62.7 (7.6)	65.5 (7.1)	67.0 (6.6)	68.3 (6.7)	<0.0001	<0.0001
Sex, %
Men	45.5	59.4	85.8	74.3	<0.0001	<0.0001
Women	54.5	40.3	14.2	25.7
Mean body mass index (SD), kg/m2	23.3 (3.5)	22.5 (3.3)	22.7 (3.9)	22.5 (3.8)	0.8955	<0.0001
Level of education, %
No formal education	20.7	22.6	21.2	27.3
Primary (1–6 years)	45.2	52.4	62.0	50.2	0.1843	<0.0001
Secondary and above	34.1	25.0	16.8	22.5
Smoking status (%)
Never smokers	68.9	47.9	8.8	20.2
Former smokers	15.7	17.4	31.0	20.9	<0.001	<0.0001
Current smokers	15.4	34.7	60.2	58.9
Mean cigarettes/day (SD)§	16.7 (12.7)	21.8 (14.8)	22.0 (16.2)	19.3 (13.1)	0.1636	<0.0001
Mean years of smoking (SD)§	33.7 (14.1)	40.2 (12.5)	42.8 (11.6)	44.0 (10.9)	0.0099	<0.0001
Mean pack-years of smoking (SD)‡	29.7 (27.2)	43.4 (30.8)	47.0 (36.3)	41.8 (28.1)	0.3776	<0.0001
Alcohol consumption
Non-drinkers	81.4	78.5	77.9	80.3
<7 drinks/week	14.2	13.9	9.7	13.8	0.2409	<0.0001
≥7 drinks/week	4.4	7.6	12.4	5.9

^*Abbreviations: SD, standard deviation.

^†Two-sided Ps comparing among 3 groups of lung cancer cases by histology type were based on analysis of variance for continuous variables or chi-square test for categorical variables.

^‡Two-sided Ps comparing total cancer cases (n = 654) with non-cancer cases were based on t test for continuous variables or chi-square test for categorical variables.

^§Among former and current smokers only.

The telomeres shortened with increasing age in both men and women of all study subjects included (Figure 1) or lung cancer cases (Supplemental Figure 1). Age explained 5.7% and 6.6% variability in telomere length for women and men, respectively. Overall the mean (standard deviation) of relative telomere length measured in 71–74 years old was 0.94 (0.20) compared with 1.18 (0.28) in 46–50 years old. Individuals in the lowest quintile of telomere length (median 0.76, range 0.19–0.83) were 5.8 years older than those in the highest quintile of telomere length (median 1.30, range 1.19–3.24) (Table 2). There were significantly higher proportions of women, higher level of education, weekly vigorous work or strenuous sports, and never smokers in the highest quintile than in the lowest quintile of telomere length. Ever smokers with the lowest quintile of telomere length smoked more numbers of cigarettes per day, years of smoking, and pack-years of smoking than their counterparts with the highest quintile of telomere length. The proportion of drinkers consuming 7 or more alcoholic beverages per week was highest in lowest quintile of telomere length.

Figure 1.

Correlation between age and relative telomere length in peripheral blood leukocytes in men and women, the Singapore Chinese Health Study 1993–2015. The red color denotes for women and the black color for men.

Table 2.

Baseline characteristics of all participants by relative telomere length The Singapore Chinese Health Study, 1993–2015

Characteristics	Total No. of Subjects	Relative telomere length in quintile [median] and (range)					Pvalues
		1st [0.76] (0.19–0.83)	2nd [0.89] (0.84–0.96)	3rd [1.00] (0.95–1.05)	4th [1.11] (1.05–1.19)	5th [1.30] (1.19–3.24)
Number of subjects	26,540	5,308	5,308	5,308	5,308	5,308
Age in years, mean (SD)	26,540	65.9 (7.8)	63.8 (7.6)	62.7 (7.5)	61.5 (7.2)	60.1 (6.9)	<0.0001
Female sex, n (%)	14,306	2,379 (44.8)	2,666 (50.2)	2,823 (53.2)	3,135 (59.1)	3,303 (62.2)	<0.0001
Level of education
No formal education, n (%)	5,525	1,195 (22.5)	1,122 (21.1)	1,086 (20.5)	1,091 (20.6)	1,031 (19.4)
Primary school, n (%)	12,032	2,528 (47.6)	2,365 (44.6)	2,408 (45.4)	2,389 (45.0)	2,342 (44.1)	<0.0001
Secondary school or higher, n (%)	8,983	1,585 (29.9)	1,821 (34.3)	1,814 (34.2)	1,828 (34.4)	1,935 (36.5)
BMI (kg/m2), mean (SD)	26,540	23.1 (3.5)	23.2 (3.5)	23.3 (3.5)	23.4 (3.5)	23.3 (3.5)	0.0008
Weekly vigorous work or strenuous sports, n (%)	4,172	803 (15.1)	797 (15.0)	818 (15.4)	858 (16.2)	896 (16.9)	0.0411
Smoking status
Never smoker, n (%)	18,021	3,200 (60.3)	3,455 (65.1)	3,611 (68.0)	3,813 (71.8)	3,942 (74.3)
Former smoker, n (%)	4,207	1129 (21.3)	925 (17.4)	802 (15.1)	709 (13.4)	642 (12.1)	<0.0001
Current smoker, n (%)	4,312	979 (18.4)	928 (17.5)	895 (16.9)	786 (14.8)	724 (13.6)
Cigarettes/day, mean (SD)†	8,519	17.6 (13.2)	16.7 (12.8)	17.1 (12.9)	16.5 (12.6)	16.4 (12.6)	0.0360
Years of smoking, mean (SD)†	8,519	36.3 (14.6)	34.7 (14.7)	34.1 (13.7)	32.5 (13.3)	31.8 (13.6)	<0.0001
Pack-years of smoking, mean (SD)†	8,519	33.6 (29.1)	30.8 (28.6)	30.5 (27.5)	28.0 (25.0)	27.9 (25.9)	<0.0001
Alcohol consumption status
Non-drinkers, n (%)	21,577	4,321 (81.4)	4,311 (81.2)	4,309 (81.2)	4,321 (81.4)	4,315 (81.3)
<7 drinks/week, n (%)	3,772	706 (13.3)	745 (14.0)	742 (14.0)	781(14.7)	798 (15.0)	0.0003
≥7 drinks/week, n (%)	1,191	281 (5.3)	252 (4.8)	257 (4.8)	206 (3.9)	195 (3.7)
Ethanol intake (g/day), mean (SD)‡	4,963	10.6 (16.5)	9.2 (15.2)	10.2 (16.9)	7.9 (13.6)	8.6 (15.0)	0.0003

^†Among former and current smokers only.

^‡Among drinkers only.

The mean relative telomere length was 1.022 (95% CI = 1.005, 1.039) in all lung cancer cases and 0.995 (95% CI = 0.992, 0.997) in non-cases (P = 0.0016) after adjustment for age, sex, education, BMI, number of cigarettes per day, number of years of smoking, and number of years since quitting smoking (for former smokers only). Longer telomeres were significantly associated with higher risk of lung adenocarcinoma after adjustment for multiple confounders or age alone (Supplemental Table 1). Compared with the lowest quintile, subjects with the highest quintile of telomere length showed a 2.8 times greater risk of developing lung adenocarcinoma (P_trend < 0.0001) after adjustment for multiple potential confounders (Table 3). Telomere length was not associated with risk of lung squamous carcinoma or other/unknown histological type of lung cancer after adjustment for age alone or multiple confounders (Supplemental Table 1, Table 3).

Table 3.

Hazard ratio (HR) of lung cancer by histological subtype in relation to relative telomere length The Singapore Chinese Health Study 1993–2015

Relative telomere length in quintile	Person-years	Adenocarcinoma		Squamous cell cancer		Cancer of other/unknown histology
		No	HR (95% CI)*	No	HR (95% CI)*	No	HR (95% CI)*
All subjects
1st (shortest)	59,803	44	1.00	28	1.00	72	1.00
2nd	62,041	50	1.29 (0.86, 1.94)	32	1.43 (0.86, 2.37)	56	0.96 (0.68, 1.37)
3rd	63,019	63	1.75 (1.18, 2.57)	19	0.96 (0.53, 1.73)	51	1.01 (0.71, 1.45)
4th	63,989	48	1.51 (1.00, 2.29)	18	1.11 (0.61, 2.02)	37	0.89 (0.59, 1.33)
5th (longest)	65,355	83	2.84 (1.94, 4.15)	16	1.13 (0.60, 2.13)	37	1.05 (0.69, 1.58)
P trend			<0.0001		0.9886		0.9714

^*Adjusted for age, sex, dialect group, education, body mass index, number of cigarettes per day, and number of years of smoking, number of years since quitting smoking (for former smokers only), and alcohol consumption.

A statistically significant, positive association between telomere length and risk of lung adenocarcinoma was present in both never (HR 3.14, 95% CI 1.80–5.49 comparing the highest with the lowest quintile, P_trend<0.0001) and ever smokers (HR 2.46, 95% CI 1.45–4.18, P_trend=0.0010) (Table 4). A similar association was seen in men (HR 2.33, 95% CI 1.46–3.74, P_trend=0.0003) and women (HR 4.26, 95% 2.11–8.61, P_trend<0.0001). When data were analyzed by the duration of follow-up, the associations between longer telomeres and higher risk of lung adenocarcinoma were comparable for the longer duration (5+ years) (HR 2.46, 95% CI 1.59–3.82, P_trend=0.0001) with the shorter duration (<5 years) of follow-up (HR 4.19, 95% CI 1.96–8.96, P_trend=0.0002). We conducted similar subgroup analyses and did not find any statistically significant association between telomere length and risk of squamous cell carcinoma or other/unknown histological types of lung cancer (all P_trend >0.50, Supplemental Tables 2 and 3).

Table 4.

Hazard ratio (HR) of lung adenocarcinoma in relation to relative telomere length by smoking status, gender and duration of follow-up The Singapore Chinese Health Study 1993–2015

Relative telomere length in quintile	Person-years	No. of adenocarcinoma	HR (95% CI)*	P for trend
Never smokers
1st (shortest)	37,950	18	1.00	<0.0001
2nd	42,057	21	1.22 (0.65, 2.29)
3rd	44,479	26	1.53 (0.84, 2.81)
4th	47,059	23	1.39 (0.74, 2.60)
5th (longest)	49,727	50	3.14 (1.80, 5.49)
Ever smokers
1st (shortest)	21,854	26	1.00	0.0010
2nd	19,984	29	1.37 (0.80, 2.32)
3rd	18,540	37	1.93 (1.16, 3.20)
4th	16,930	25	1.63 (0.93, 2.85)
5th (longest)	15,628	33	2.46 (1.45, 4.18)
Males
1st (shortest)	31,406	34	1.00	0.0003
2nd	29,425	29	1.04 (0.63, 1.70)
3rd	27,902	39	1.55 (0.97, 2.47)
4th	25,261	28	1.39 (0.83, 2.30)
5th (longest)	23,602	41	2.33 (1.46, 3.74)
Females
1st (shortest)	28,397	10	1.00	<0.0001
2nd	32,616	21	2.09 (0.98, 4.44)
3rd	35,117	24	2.37 (1.13, 4.96)
4th	38,728	20	1.99 (0.93, 4.29)
5th (longest)	41,753	42	4.26 (2.11, 8.61)
<5 years of follow-up
1st (shortest)	25,007	10	1.00	0.0002
2nd	25,399	14	1.64 (0.73, 3.69)
3rd	25,478	18	2.37 (1.09, 5.16)
4th	25,561	13	2.00 (0.87, 4.62)
5th (longest)	25,611	24	4.19 (1.96, 8.96)
≥5 years of follow-up
1st (shortest)	34,796	34	1.00	0.0001
2nd	36,641	36	1.19 (0.74, 1.90)
3rd	37,541	45	1.56 (1.00, 2.45)
4th	38,428	35	1.36 (0.84, 2.20)
5th (longest)	39,744	59	2.46 (1.59, 3.82)

^*Adjusted for age, sex, dialect group, education, body mass index, number of cigarettes per day, and number of years of smoking, number of years since quitting smoking (for former smokers only), and alcohol consumption.

The cumulative incidence rates of lung adenocarcinoma are shown in Figure 2 for ever smokers and never smokers separately with different levels of telomere length after taking into account the competing risk of death from other causes. Higher incidence rates of lung adenocarcinoma were seen in both never and ever smokers with longer telomeres compared to those with shorter telomeres. For example, the cumulative incidence rates of lung adenocarcinoma were 0.79% (95% CI 0.38–1.21%), 1.08% (95% CI 0.71–1.45%) and 2.16% (95% CI 1.46–2.87%), respectively, in never smokers with the lowest, intermediate (2^nd–4^th quintile) and highest quintile of telomere length by the age of 80 years. The corresponding figures in ever smokers were at 1.15% (95% CI 0.59–1.72%), 2.55% (95% CI 1.95–3.15%), and 4.12% (95% CI 2.67–5.58%).

Figure 2.

Cumulative incidence rate of lung adenocarcinoma in never and ever smokers by categories of relative telomere length in peripheral blood leukocytes after adjustment for competing risk of death in the Singapore Chinese Health 1993–2015. The green color denotes for the short telomeres (Q1: the lowest quintile of telomere length), the purple color for the intermediante telomeres (Q2–Q4: the 2^nd–4^th quintile of telomere length), and the red color for the long telomeres (Q5: the highest quintile of telomere length). The solid lines denote for ever smokers and dask lines for never smokers.

DISCUSSION

In this population-based prospective cohort study of 26,540 individuals with an average follow-up of 12 years, we showed that longer telomeres in peripheral blood leukocytes were associated with significantly increased risk of developing lung adenocarcinoma, but not with squamous cell carcinoma or other histological types of cancer. This positive association between telomere length and lung adenocarcinoma risk was present in sub-populations by smoking status, gender and the duration of follow-up. Our results are consistent with findings from Mendelian randomization studies.^11–13 It is worth noting that two of these reports were based on virtually the same study subjects.^{12, 13}

Is telomere length in peripheral blood leukocytes a valid surrogate marker for target tissues such as the lung? Telomere length varies across different somatic tissues dependent on their replicative activities. Telomere length in peripheral blood leukocytes and in the lung and skin tissues is usually shorter than those in the muscle and fat tissues because the former have higher proliferating potential than the latter.^{22, 23} Within a given individual, telomere lengths in different organs and tissue types are highly correlated with each other. Several human studies have showed that telomere length in peripheral blood leukocytes was positively correlated with telomere length in buccal cells,²⁴ fibroblasts,²⁴ skin,^{25, 26} and synovial membrane.²⁵ One study showed a high correlation for telomere length in peripheral blood leukocytes with the muscle cells (r²=0.71), fat (r²=0.69), and skin (r² =0.69) (all Ps <0.0001).²³ Furthermore, telomere length in both skin and skeletal muscle cells was highly correlated with that in the lung tissue (r² ≥ 0.77) after controlling for chronological age.²² These results strongly suggest that peripheral blood leukocytes are good surrogates for lung and other tissue types in the measurement of telomere length.

Several epidemiological studies have examined and reported inconsistent associations between peripheral blood leukocyte telomere length and lung cancer risk. One of the chief reasons for these inconsistencies may stem from study design. Retrospective case-control studies often reported an increased risk of lung cancer with shorter telomeres.^27–29 Telomere shorting may occur after diagnosis due to treatment or disease progression.³⁰ A recent study reported significantly shorter telomeres in patients with advanced stage of lung cancer than those with early stage lung cancer, suggesting that tumor progression may impact peripheral blood leukocyte telomere length.²⁹ Thus retrospective studies, in which blood samples were collected from patients after cancer diagnosis or treatment, would yield shorter telomeres than those in pre-diagnosed blood samples, resulting in a spurious elevation in risk of lung cancer associated with shorter telomeres.

Prospective studies, in which blood samples are collected from study subjects before their cancer diagnosis, avoid such a shortcoming. There have been two prospective cohort studies on telomere length and risk of total or lung cancer incidence. The earlier study reported a two-fold increased risk of total cancer incidence for the shortest third compared with the longest third of peripheral blood leukocyte telomeres, but that study suffered from a very small sample size with only 787 total subjects and 92 total cancer cases.³¹ The later cohort study in a general population in Copenhagen with a large sample size did not find a statistically significant association between peripheral blood leukocyte telomere lengths and risk of total cancer or lung cancer incidence.^{32, 33} However all these analyses did not separate cases of adenocarcinoma from squamous cell carcinoma or other histological types of lung cancer. In addition, three case-control studies of lung cancer nested in the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study,³⁴ the Shanghai Women’s Health Study,³⁵ and the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial³⁶ all reported a significantly positive association between baseline peripheral blood leukocyte telomere length and risk of lung cancer. In the pooled analysis of these 3 case-control studies, Seow et al reported a statistically significant increased risk of lung adenocarcinoma, but not squamous cell carcinoma, in individuals with longer peripheral blood leukocyte telomeres.³⁶ Our results are consistent with this pooled analysis of 3 prospective case-control studies.

Studies using genetic variants as proxy measures for telomere length could avoid the biases due to disease progression and treatment, and other attributes to telomere shortening, such as aging and oxidative stress, since genetic variants are unchanged by these factors. A summed genetic score of various single genetic polymorphisms in TERT, TERC and other candidate genes for longer telomeres, identified in a genome-wide association study,³⁷ was significantly associated with increased risk of lung cancer among never-smoking Asian women¹¹ and increased risk of lung cancer and melanoma in the Copenhagen cohort study.³³ More recently, two separate reports based on the Mendelian randomization study of the same genome-wide association studies of lung cancer reported that genetic allele scores for longer telomeres were significantly associated with increased risk of lung adenocarcinoma, but was not associated with squamous cell carcinoma.^{12, 13} Consistent with those of previous studies, our prospective cohort study in a single ethnic population further support the results that longer peripheral blood leukocyte telomeres are associated with increased risk of lung adenocarcinoma.

The biological mechanism linking longer telomeres to lung carcinogenesis is unclear. Shorter telomeres may act as tumor suppressors³⁸ that can protect against carcinogenesis by triggering programmed cell death in the presence of functional cell cycle checkpoints and intact apoptotic pathways.³⁹ In contrast, cells with longer telomeres have higher proliferative capacity. Each round of genome replication has the potential to introduce genetic mutations and chromosomal alterations, which may promote malignant transformation.⁴⁰ Alternatively peripheral blood leukocyte telomere length may serve as an indicator of other factors for risk of lung cancer. Clinical studies have found a statistically significant association between telomere length and higher percentage of regulatory T cells (T_reg) in the blood from patients with hepatocellular or renal cell carcinoma.^{41, 42} High levels of T_reg have been detected in various cancers and T_reg are proposed to play an important role in immune suppression, thereby establishing tumor immune tolerance.⁴³ The proportion of T_reg was significantly higher in bronchoalveolar lavage fluid from patients with adenocarcinoma than those with squamous cell carcinoma.⁴⁴ This may contribute to the observed positive association for peripheral blood leukocyte telomere length with the risk of lung adenocarcinoma, but a null association with the risk of squamous cell carcinoma. Future studies are warranted to clarify the association between peripheral blood leukocyte telomeres and risk of subtype lung cancer.

The present study has several strengths. The prospective design minimized the potential impact of progression and treatment of lung cancer on telomere length, since telomere length was determined many years before diagnosis of lung cancer. The quantification of telomere length for the entire cohort allowed for the estimation of cumulative absolute risk of lung adenocarcinoma after taking into account for competing risk of death, minimizing the potential for inflating the estimate of cumulative incidence risk.²¹ The direct quantification of telomere length in a homogenous Chinese population eliminated potential confounding by pleiotropy, population stratification, and ancestry that are present in the Mendelian randomization studies. The long-term and complete follow-up further reduced potential bias due to the impact of undiagnosed lung cancer on telomere length. The relatively large number of lung cancer cases provided sufficient statistical power to yield robust risk estimates of lung adenocarcinoma in sub-population by smoking status, gender and duration of follow-up. Our study also has some limitations. As with any observational studies, our results are subject to residual confounding by both measured and unmeasured confounders in the study population. Another limitation is whether the telomere length in peripheral blood leukocytes reflects the telomere length in the lung, although previous studies have shown high correlation of telomere length measures between the two tissue types.^{22, 23} The relatively small number of lung squamous cell carcinoma may be one possible reason for a null association with telomere length.

In summary, the present study demonstrates a dose-dependent association for telomere length in peripheral blood leukocytes at baseline with increased risk of lung adenocarcinoma. Our findings and those from previous prospective studies and Mendelian randomization studies support a potential etiological role of telomere length in the development of lung adenocarcinoma. Future research efforts need to be undertaken to elucidate the biological mechanism for telomeres in the development of lung adenocarcinoma.

NOVELTY AND IMPACT.

The role of telomere length in the development of cancer is intriguing, given the inconsistent results from early retrospective observational studies and more recent prospective and Mendelian randomization studies. This prospective cohort study demonstrates a robust, positive association between direct measurement of telomere length in peripheral blood leukocytes and the risk of developing lung adenocarcinoma. Consistent with results of recent prospective and Mendelian randomization studies, the findings of the present study support a potential etiological role of telomere length in the development of lung adenocarcinoma.