Causality between telomere length and breast diseases: a two-sample bidirectional Mendelian randomization study

Abstract

Background:

The relationship between telomere length and breast diseases remains unclear, with conflicting evidence for breast cancer. Using an innovative genetic approach, we were the first to comprehensively assess their bidirectional causal relationship.

Methods:

Telomere length, breast cancer, benign neoplasm of breast, and breast inflammation were extracted from the genome-wide Association study (GWAS) database as the basis for large-scale population studies. The interaction of telomere length and breast diseases as exposure and outcome factors was analyzed by Mendelian randomization (MR).

Results:

When telomere length was used as an exposure factor and breast diseases as an outcome, the P value of MR was less than 0.05. Breast cancer (odds ratio (OR) = 1.130, 95% confidence interval (CI) = 1.047–1.219, P = 0.0016), benign neoplasm of breast (OR = 1.002, 95%CI = 1.001–1.004, P = 0.0007) and breast inflammation (OR = 1.487, 95%CI = 1.008–2.191, P = 0.0453). When breast diseases were taken as an exposure factor and telomere length was taken as an outcome, the P value of MR between breast cancer, benign neoplasm of breast, and telomere length was greater than 0.05, and breast inflammation could not be calculated by MR.

Conclusion:

Telomere length is a risk factor for breast diseases, and longer telomeres increase the risk of breast cancer, benign neoplasm of breast, and breast inflammation. However, the reverse study showed no causal association between breast cancer, benign neoplasm of breast and telomere length, and the causal association between breast inflammation and telomere length was not clear. Moreover, further studies are needed to validate our findings in non-European populations.

Introduction

Telomeres are protein-DNA complexes, and human telomeres consist of a complex between telomere DNA and a six-protein complex called asylum, which protects chromosome ends from illegal joining and excising^[1,2]. Telomeres shorten continuously as cells divide, and short dysfunctional telomeres signal DNA damage, which triggers cellular responses that lead to aging or apoptosis[3].

HIGHLIGHTS

• This study is the first large-scale Mendelian randomization (MR) analysis to explore the bidirectional causal relationship between telomere length and breast diseases, including breast cancer, benign neoplasm of the breast, and breast inflammation.

• Longer telomeres increase the risk of breast cancer, benign breast neoplasms, and breast inflammation, but there is no evidence for a reverse causal relationship.

• Findings align with previous research showing a paradoxical association between telomere length and cancer risk, supporting the clonal amplification model and highlighting the role of POT1 mutations.

Changes in telomere length have been shown to be associated with a variety of benign and malignant diseases, such as goiter, thyroid cancer, uterine fibroids, pancreatic cancer, idiopathic pulmonary fibrosis, and chronic obstructive pulmonary disease^[4-6]. However, the relationship between telomere length and breast diseases is still not fully clarified. In clinical practice, breast cancer, benign neoplasm of breast and breast inflammation are the three most common breast diseases. Among them, the incidence of benign breast diseases is higher[7]. The causal link between telomere length and breast cancer remains debated^[8,9]. While earlier studies have indicated an association with prognosis, the evidence is limited[10]. No studies have shown a causal link between telomere length and benign breast diseases. Therefore, it is necessary to further study the relationship between telomere length and breast diseases.

Our study, using Mendelian randomization (MR) methods, comprehensively explored the bidirectional causal association between telomere length and breast diseases. MR uses genetic variation as an instrumental variable for modifiable risk factors affecting population health to estimate the causal effect of exposure on outcomes[11]. Genetic variants that are randomly assigned at conception, are not influenced by behavioral or environmental factors, and are less susceptible to bias by reverse causation can overcome the unmeasured confounding and reverse causation problems typical of traditional observational epidemiology^[12,13]. Therefore, it is very important to use MR methods to explore the relationship between telomere length and breast diseases. As all participants were of European ancestry, the generalizability of our findings to other populations may be limited.

Methods

Study population

The single nucleotide polymorphisms (SNPs) associated with exposure and outcome were all from the GWAS database, and the subjects were all European. The study population for telomere length was derived from the “ieu-b-4879” dataset (published in 2021), which contained 472 174 survey participants covering 20 134 421 SNPs. The breast cancer study population was drawn from the “ieu-a-1126” dataset (published in 2017) of Breast Cancer Association Consortium (BCAC), which contained 228 951 survey respondents covering 10 680 257 SNPs. The study population for benign neoplasm of breast used the UKBiobank “ukb-b-8549” dataset (published in 2018), which contained 463 010 survey participants and covered 9 851 867 SNPs. The breast inflammation study population was derived from the “finn-b-N14_INFLAMMBREAST” dataset (published in 2021), which contained 115 787 survey respondents covering 16 379 554 SNPs. All of the above data (Table 1) were publicly published and participants gave informed consent to participate and follow up. All the GWAS datasets used in this study included adjustments for key confounders. Specifically, the analyses adjusted for age and population stratification via principal component analysis. These adjustments help minimize confounding and ensure the selected genetic variants satisfy the assumptions of MR.

Table 1.

Data on single nucleotide polymorphisms associated with exposure and outcome

	Time	Dataset	Sample size	SNPs number
Telomere length	2021	ieu-b-4879	472 174	20 134 421
Breast cancer	2017	ieu-a-1126	228 951	9 851 867
Benign neoplasm of breast	2018	ukb-b-8549	463 010	9 851 867
Breast inflammation	2021	finn-b-N14_INFLAMMBREAST	115 787	16 379 554

Acquisition of instrumental variables

Instrumental variables should satisfy the following three assumptions (Fig. 1): (1) Genetic variation must be closely related to exposure factors (P < 5 × 10^‒8); (2) Genetic variation cannot be associated with any confounding factors; (3) Genetic variation can only affect the outcome through exposure factors, but not through other ways.

Figure 1.

Diagram of assumptions that tool variables need to satisfy.

We first extracted the SNPs of the exposure factors, and in order to ensure the independence of the selected genetic variants, we removed linkage disequilibrium (LD) by setting the parameters r² = 0.001 and kb = 10 000. Then SNPs of the three outcome variables were extracted based on the exposure factors. Finally, the direction of effects between exposure and outcome associations was coordinated, and SNPs with incompatible alleles and intermediate allelic frequency palindromes were removed.

Instrument strength assessment

To evaluate the strength of the genetic instruments for telomere length (exposure), we calculated the F-statistic for each SNP using the formula $<tex-math><![CDATA[\$F=\frac{{\beta }^2}{S{E}^2},\$]]></tex-math>$ and the proportion of variance explained (R²) was estimated based on the SNP effect sizes, allele frequencies, and sample size. The total R² was obtained by summing the individual SNP R² values. Instruments with F-statistics >10 are generally considered strong and less likely to suffer from weak instrument bias.

Mendelian randomization analysis

Five methods[14] [inverse variance weighting (IVW), MR Egger, weighted median, simple mode, and weighted mode] were used to evaluate the causal association between telomere length and breast diseases. The IVW results were the standard and the results of other methods were used as reference. IVW is a method of aggregating two or more random variables to minimize the variance of a sum in which the weighting of each random variable in the sum is inversely proportional to its variance, which is often used to combine the results of independent studies[15]. In addition, we will use the above methods to reverse evaluate the causal association between breast diseases and telomere length, that is, breast diseases as exposure factors and telomere length as an outcome variable.

Sensitivity analysis

The IVW and MR-Egger methods were used for the heterogeneity test, and a P value <0.05 indicated that there was heterogeneity in the study[10]. We performed a horizontal pleiotropy test to assess the stability of the results. The pooled effects of residual SNPs were evaluated using IVW and MR-Egger methods. If the combined effect is consistent with the main effect, it indicates that no single SNP has an excessive influence on the MR Analysis[10].

Statistics and plotting

We performed data extraction, MR analysis, sensitivity analysis and plotting on R (version 4.3.3) through packages of TwoSampleMR and forestploter. MR results were presented using OR and corresponding 95%CI, and bilateral P <0.05 was considered statistically significant.

Results

Instrument strength

A total of 154 independent SNPs (P < 5 × 10^‒8, LD r² < 0.001, clumping distance = 10 000 kb) were selected as instrumental variables for telomere length from the GWAS dataset ieu-b-4879 (Supplemental Digital Content, File 1, available at: http://links.lww.com/MS9/A888). The mean F-statistic across all SNPs was 115.74, indicating strong instrument strength. The total variance in telomere length explained by these SNPs was 3.77%, suggesting the instruments provide reasonable predictive power for the exposure.

Instrumental variables

Breast cancer obtained 137 SNPs, SNPs that removed incompatible allele (rs9940099) and SNPs that palindromed with intermediate alleles (rs2306646, rs56178008, rs670180). A total of 133 SNPs were left for subsequent analysis. The benign neoplasm of breast obtained 58 SNPs, SNPs without incompatible alleles, and SNPs with intermediate allelic frequency indromes removed (rs2306646, rs56178008, rs670180). Fifty-five SNPs were left for subsequent analysis. Breast inflammation received 142 SNPs. The SNPs with incompatible allele (rs9940099) and those with intermediate allelic frequencies (rs2276182, rs2306646, rs56178008, rs670180) were removed, leaving 137 SNPs for subsequent analysis.

Mendelian randomization

IVW method showed that telomere length was associated with breast cancer (OR = 1.130, 95%CI = 1.047–1.219, P = 0.0016), benign neoplasm of breast (OR = 1.002, 95%CI = 1.001–1.004, P = 0.0007) and breast inflammation (OR = 1.487, 95%CI = 1.008–2.191, P = 0.0453). Although in breast inflammation, except IVW method, the P values of the other four results were ≥0.05, their OR values were all >1, in the same direction as IVW method, so the causal association between telomere length and breast inflammation was established. Specific results are shown in Fig. 2. The results of reverse causality showed that the MR results of breast cancer, benign neoplasm of breast and telomere length were P >0.05 (IVW method), indicating that breast cancer and benign neoplasm of breast were not causally associated with telomere length when they were used as exposure factors. However, when breast inflammation was used as the exposure and telomere length as the outcome, valid MR estimates could not be obtained, possibly due to an insufficient number of effective SNPs after harmonization or potential bias from weak instruments.

Figure 2.

Forest plot of MR analyses of the effect of telomere length on breast diseases.

SE, standard error; OR, odds ratio.

Sensitivity analysis

The results of heterogeneity tests (Table 2) showed that the MR Egger test (Q = 324.551, P = 1.968573e-18) and IVW test (Q = 325.208, P = 2.558354e-18) for breast cancer indicated significant heterogeneity. This suggests variability in the SNP-specific causal estimates, which may arise from differences in linkage disequilibrium, population structure, or potential pleiotropic effects of some SNPs. However, the horizontal pleiotropy test and MR Egger intercept test (Table 3) showed no evidence of directional pleiotropy (all P values > 0.05), and leave-one-out sensitivity analysis confirmed that no single SNP disproportionately influenced the MR estimates. Furthermore, consistent effect directions across multiple MR methods, including MR Egger and weighted median approaches, support the robustness of the causal inference.In contrast, the MR Egger test (Q = 50.217, P = 0.583) and IVW test (Q = 50.266, P = 0.619) for benign neoplasm of breast, and the MR Egger test (Q = 132.392, P = 0.547) and IVW test (Q = 132.857, P = 0.560) for breast inflammation showed no significant heterogeneity.

Table 2.

Results of heterogeneity test in sensitivity analysis

	Q	Q_df	P value
Breast cancer
MR Egger	324.551	131	1.968573e-18
IVW	325.208	132	2.558354e-18
Benign neoplasm of breast
MR Egger	50.217	53	0.583
IVW	50.266	54	0.619
Breast inflammation
MR Egger	132.392	135	0.547
IVW	132.392	136	0.547

Q, residual sum of squares; Q_df, the degrees of freedom of residual sum of squares.

Table 3.

Results of horizontal pleiotropy test in sensitivity analysis

	Egger intercept	SE	P value
Breast cancer	−0.001	0.002	0.607
Benign neoplasm of breast	7.206932e-06	3.241614e-05	0.825
Breast inflammation	0.007	0.010	0.497

SE, standard error.

Discussion

To our knowledge, this is the first large-scale MR study to comprehensively explore the bi-directional causal relationship between telomere length and breast cancer, benign neoplasm of breast, and breast inflammation. Previously, MR has also been used to explore women’s sleep characteristics[16], physical activity and sedentary time[17], and circulating lipids[18] in relation to breast cancer. Previous studies have shown conflicting results regarding the causal association between telomere length and breast cancer risk. One study suggested that longer telomeres may increase the risk of breast cancer[8], while another found no evidence of a causal relationship[9]. These inconsistencies highlight the need for further investigation using robust methodologies such as MR. The relationship between telomere length and breast cancer prognosis has been explored in previous study, which reported an inverse association between telomere length and prognosis, particularly among estrogen receptor negative patients[10]. Nevertheless, their data were limited, and the datasets and analysis angles were different from ours. Moreover, there is no MR study on telomere length and benign breast diseases.

Our findings demonstrate a causal association between telomere length and three breast-related conditions: breast cancer, benign neoplasm of the breast, and breast inflammation, suggesting that longer telomere length may be a potential risk factor for these diseases. Although significant heterogeneity was detected in the breast cancer MR analysis, the consistency across various MR methods and the absence of directional pleiotropy support the robustness of our findings. Nonetheless, this heterogeneity highlights the need for cautious interpretation and further validation in future studies. However, reverse MR analysis did not support a causal effect of breast cancer or benign neoplasm of the breast on telomere length. The reverse association between breast inflammation and telomere length remains inconclusive. Although we identified a forward causal relationship between telomere length and breast inflammation, we were unable to obtain valid estimates in the reverse direction. This may be due to an insufficient number of robust instrumental variables for breast inflammation, or potential issues such as weak instrument bias or collider bias. Future research utilizing larger and more refined GWAS datasets is needed to clarify this relationship.

Previous studies have indicated that changes in telomere length are associated with a variety of benign and malignant diseases^[4-6]. However, the relationship between telomere length and cancer seems paradoxical, with individuals with short telomeres having an increased risk of cancer because short telomeres lead to genomic instability and apoptosis, while individuals with long telomeres have an increased risk of major cancers too^[19–26]. The two-stage clonal amplification model proposed by Abraham Aviv et al solves this paradox well[19]. Their study[19] suggests that genomic instability in single-gene diseases with short telomeres is constitutive, systemic, and already present at the stem cell level, and is not consistent with the general population’s susceptibility to cancer, because cancer-associated genomic instability in the general population is acquired at the clonal level, which mainly occurs at specific sites undergoing clonal amplification. Emily A DeBoy et al’s study[4] showed that people with POT1 gene mutations had longer telomeres than their non-carrier relatives, thus demonstrating that POT1 mutations associated with long telomere length predisposes familial clonal hematopoietic syndrome, which is associated with a range of benign and malignant solid tumors. Other studies have also suggested that precancerous cells can avoid aging and continue to proliferate by upregulating telomerase or lengthening telomeres by alternative mechanisms[27], which in part explains the increased risk of cancer in individuals with long telomeres.

In a previous study on telomere length and breast cancer, Beatriz Martinez-Delgado et al found that telomere shortening was associated with earlier ages of breast cancer onset in successive generations by surveying 623 families with breast cancer[28]. In A population-based telomere study of 1995 breast cancer cases and 2296 controls in Poland, no significant association was found between a single SNP or haplotype and breast cancer risk. But a small number of SNPs may be associated with a reduced risk of breast cancer in individuals with a family history of breast cancer[29]. Andrew J Pellatt et al, by evaluating breast cancer health disparities among non-Hispanic white and Hispanic and Mexican women in the United States, noted that longer telomere length is associated with an increased risk of breast cancer, which is consistent with our findings[30]. However, the relationship between telomere length and benign breast diseases has not been studied, and the mechanism of their relationship is not clear. Studies have found that there is a relationship between benign breast diseases and breast cancer risk, and that the risk varies by histological category and hormonal status of benign breast diseases^[31,32]. Therefore, we speculate that the causal association between telomere length and benign breast diseases may be related to the relationship between benign breast diseases and breast cancer.

Although our research has many advantages. For example, this is the first time that the relationship between telomere length and breast diseases has been comprehensively explored in a large population, and the characteristics of the MR method also avoid the influence of confounding factors. However, the study has several limitations. First, all participants included in the GWAS datasets were of European ancestry. While this helps reduce the risk of population stratification within the analyses, it also limits the generalizability of our findings to other ethnic groups. Future research involving more diverse populations is essential to validate our results and assess whether the observed causal effects hold across different ancestral backgrounds. Second, although we identified a causal association between telomere length and all three breast conditions, the effect size for benign neoplasm of the breast was very small (OR = 1.002). The clinical or biological relevance of such a modest effect remains uncertain and warrants further investigation. Third, while our study demonstrates telomere length as a potential risk factor for breast disease occurrence, the underlying molecular mechanisms remain unclear – especially for benign breast conditions, where supporting mechanistic evidence is currently lacking. Further experimental and translational studies are needed to elucidate these biological pathways. Fourth, this study is a retrospective study, and further prospective studies should be conducted to obtain stronger data support and evidence.

Conclusion

Our findings provide novel insights into the role of telomere length in breast diseases. We demonstrated that longer telomeres are causally associated with an increased risk of breast cancer, benign neoplasm of the breast, and breast inflammation. Reverse MR analysis showed no evidence of causal effects in the opposite direction for breast cancer and benign breast neoplasms, while the association for breast inflammation remains inconclusive due to technical limitations. These results suggest that telomere length may serve as a potential biomarker for the early identification of individuals at higher risk for various breast diseases. Understanding the biological mechanisms underlying these associations may help guide personalized prevention strategies and facilitate the development of telomere-targeted interventions. Future studies should focus on elucidating the molecular pathways involved and validating these findings across diverse populations to enhance clinical translation.

Ethical approval and consent to participate

This study is based on publicly available summary-level data from genome-wide association studies (GWAS), and no individual-level data were used. The original studies providing these data had obtained ethical approval from their respective institutional review boards (IRBs) and obtained informed consent from all participants. Therefore, no additional ethical approval or consent was required for this study.

Sources of funding

This study was supported by National Natural Science Foundation of China Youth Science Fund Project (No. 82203141), Basic and Applied Basic Research Program of Guangzhou City University (Institute), and Enterprise Joint Funded Project (No. 2023A03J0720).

Author contributions

S.T. designed the experimental method and completed the statistics and plotting. S.L., J.C., and Y.Y.C. reviewed the literature. S.L. finished the first draft of the manuscript. S.T. and L.X.X.D. revised the manuscript.

Conflicts of interest disclosure

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Guarantor

Song Tang.

Research registration unique identifying number (UIN)

Not applicable.

Provenance and peer review

Not commissioned, externally peer-reviewed.

Data availability statement

The datasets used and analyzed during the current study are available at IEU OpenGWAS project (mrcieu.ac.uk).

Acknowledgements

We thank those people and staff who contributed data to GWAS.