Long-Term Impact of Stressful Life Events on Breast Cancer Risk: A 36-Year Genetically Informed Prospective Study in the Finnish Twin Cohort

INTRODUCTION

Breast cancer remains the predominant and lethal form of malignancy affecting women worldwide, with 2.3 million new cases diagnosed globally in 2020 alone [1]. Twin studies reveal that 31% of variability in risk is attributed to genetic factors [2], encompassing well-known high-impact genes like BRCA1/2, CHECK2, and PALB2 [3], as well as numerous genes with single nucleotide polymorphisms (SNPs), collectively quantifiable in a polygenic risk score (PRS) [4]. While significant progress has been made in understanding the genetic underpinnings of breast cancer, the full picture of its aetiology remains a work in progress. Environmental factors and health-related behaviours, including established ones such as alcohol consumption, obesity, female reproductive factors, and hormone exposures, contribute significantly to breast cancer risk [5].

Stress – the physical and emotional response to challenging or demanding situations– has potential implications for later life health through various physiological mechanisms. Chronic or repeated stress disrupts the hypothalamic-pituitary-adrenal (HPA) axis which can lead to elevated cortisol levels, a hormone essential in mammary gland function and implicated in promoting cell proliferation and tumour growth [6]. Stress weakens the immune system, hindering its ability to eliminate abnormal cells via immune evasion or changes in DNA repair mechanisms [7], and is possibly reflected in epigenetic profiles [8].

The link between stressful life events and breast cancer risk is a complex one, with ongoing epidemiological research exploring the potential connection yielding diverse results. While some studies suggest a positive association between stressful life events and breast cancer risk, others, such as the UK study of 106000 women [9] report no consistent evidence for an association of breast cancer risk with perceived stress levels or adverse life events in the preceding 5 years. A 2019 meta-analysis, drawing data from prospective and retrospective cohort studies assessing the impact of stress, or lack thereof, on breast cancer incidence published before 2018, provided a broader perspective highlighting a moderate overall association between stressful life events and breast cancer risk [(pooled risk ratio: 1.11 (95%CI 1.03-1.19)] [10]. A previous publication by Lillberg et al. (2003) on the Finnish Twin Cohort demonstrated an association between stressful life events and a heightened risk of incident breast cancer [11].

There is also a familial and genetic basis to many stressful life events, reflecting the biological basis of reactions to external stressors [12, 13]. Thus, a possible explanation for the association between stressful life events and breast cancer risk may be shared genetic predisposition. This hypothesis can be evaluated by examining monozygotic (MZ) twin pairs, in which the twin sisters differ both in their experience of stressful life events and breast cancer occurrence.

Alternatively, a PRS can be constructed, based on numerous low-impact genetic variants across the genome identified from genome-wide association studies, to estimate inherent susceptibility to breast cancer. By examining the associations of breast cancer PRS with stressful life events, presence or absence of shared genetic predisposition can be tested.

The association between stressful life events years or decades prior to breast cancer diagnosis may be reflected in changes in DNA methylation. DNA methylation is a dynamic epigenetic modification, acting thus as a potential mediator between genotypes, environmental stressors and gene expression. Exposures such as smoking and experiences such as stressful life events encountered throughout the lifespan, can induce significant epigenetic variation through mechanisms such as altered DNA methylation patterns and chromatin remodelling [14]. This emerging understanding of epigenetic plasticity holds relevance in the context of stress and breast cancer. Adverse experiences have been implicated in modulating these epigenetic pathways within breast tissue, potentially leading to altered gene expression profiles that favour tumorigenesis [15, 16]. In Bode et al. (2024) [17] it was shown that 212DNA methylation sites are associated specifically with overall environmental risk for breast cancer on average 11 years prior to diagnosis among Finnish twin pairs discordant for breast cancer.

Our current investigation extends the study by Lillberg et al (2003) by 22 years, spanning 36 years to examine the association of the number of stressful life events with breast cancer risk. The analysis includes a comprehensive assessment of established and proposed risk factors as well as analysis of genetic and epigenetic information. We conducted standard cohort analyses followed by within-family analyses in MZ and dizygotic (DZ) twin pairs concordant and discordant for breast cancer occurrence. Leveraging the extensive range of birth years within our cohort, the analysis was further enriched by stratification by birth decade. To deepen our understanding of the association between stressful life events and breast cancer risk, we employed a multifaceted approach, whereby PRS and DNA methylation patterns were examined to evaluate the relationships between stressful life events, genetic predisposition and epigenetic methylation marks and breast cancer incidence.

Identifying DNA methylation sites associated with both environmental breast cancer risk and exposure to stressful life events in our cohort could provide crucial insights into how stress experiences might influence breast cancer risk through long-term epigenetic modifications. By analysing PRS and DNA methylation patterns alongside exposure to stressful life events, this study aims to provide a more holistic view on breast cancer risk.

METHODS

RESULTS

Environmental Breast Cancer Risk related DNA methylation

To assess if environmental breast cancer risk related DNAm is associated with stressful life events, we followed up our previous findings on 212 such CpG sites (Bode et al. 2024, [17]) in the 319 pairs with stressful life event score and sampled for DNAm analysis, on average, 16 years prior to breast cancer diagnosis. Their mean age at the 1981 questionnaire study was 40.0 (SD 9.8) years, and mean age at DNA sampling was 63.2 (8.9) years; these ages did not differ by zygosity. The intraclass correlation for the stressful life events score was 0.44 (95%CI 0.31-0.55) in MZ pairs indicating substantial non-genetic influence.

Our analysis implicated 42 of these previously identified CpG sites, with significant associations between stressful life events exposure and breast cancer risk related DNAm. These CpG sites are in 32 known genes (Supplemental File 1). In total, 197/212 CpG sites displayed association trend with stressful life events in the same direction as with breast cancer risk reported by Bode et al. (2024)[17] (Supplemental File 1). This significant skew towards hypomethylation associated with stressful life events exposure (binomial test p = 1.19*10^-41 is in line with our previous findings that showed that hypomethylation associates with an exposure to environmental breast cancer risk factors (Bode et al. 2024, [17]) (Figure 1).

• Download figure
• Open in new tab

Figure 1: Association between stressful life events exposure and DNAm differences among all pairs

On the y-axis the effects of exposure to stressful life events (SLE) on breast cancer related differences in DNAm in twin pairs (172 DZ and 147 MZ pairs) for 212 CpG sites identified in Bode et al. (2024) [17]. On the x-axis the hazard ratios (HR) of breast cancer (BC) in relation to changes in DNAm according to Bode et al. (2024) [17]. The validated data points indicate that the association between stressful life events and DNAm have a p-value < 0.05.

➔ This Figure was created with R version 4.3.2.

To further isolate the influence of germline genetics on the observed effects of stressful life events on DNAm, we repeated the analysis solely on the 174 MZ pairs across all 212 DNAm sites (Supplemental File 1). The effect sizes of the within pair difference in stressful life events on the within pair difference in DNAm at the aforementioned 42 CpG sites were 36% larger in MZ pairs compared to DZ pairs (paired t-test, p = 0.012). This suggests that the observed associations between exposure to stressful life events and DNAm are largely independent of germline genetic factors. Further, within the MZ twin pairs the effect direction in 193/212 CpG sites was in line with that observed in Bode et al. (2024) [17] for breast cancer risk (binomial-test p = 1.9*10^-37).

As a sensitivity test, we repeated the analyses in the Older Finnish Twin Cohort with 154 female twin pairs (96 MZ and 58 DZ pairs) who remained cancer-free until the end of the follow-up in 2018, to exclude the potential effect of future cancer diagnosis on the association between DNAm at the 212 CpG sites and stressful life events. A total of 174 of the 212 CpG sites (binomial test, p = 1.1*10^-20) showed directional consistency with Bode et al. (2024) [17] (Supplemental Table 1). We further restricted the analysis to the 96 MZ twin pairs (Supplemental Table 1), leveraging their matched germline genetics despite the smaller sample size. Even within this subset, 182 out of 212 CpG sites showed directional consistency with Bode et al. (2024) [17] (binomial test, p = 9.8*10^-28). These findings strengthen the evidence that DNAm patterns associated with environmental breast cancer risk might also be linked to stressful life events exposure, independent of both germline genetics and future breast cancer diagnosis.

DISCUSSION

We conducted a genetically informed study on the impact of stressful life events on breast cancer incidence with 36 years of follow-up. Our results emphasise the long-term association between stressful life events and breast cancer incidence, while pointing towards the limited role of genetic risk, as assessed by PRS and within-pair analyses of MZ twin pairs. Instead, by replicating environmentally driven breast cancer risk -related DNA methylation patterns, our study suggests that DNA methylation could play a role in mediating the effects of stressful life events on breast cancer incidence. Intriguingly, our analysis also revealed a birth-decade difference in risk factor association, with the HR of stressful life events varying across generations, suggesting a stronger association for those born before than after 1950.

It is well established that the balance of neuroendocrine hormones in women is easily affected by psychological trauma [22], though the impact of stressful life events on health outcomes, especially cancer, is far from fully understood. Major traumatic events are potent sources of stress and more easily attributable to adverse health events later in life [23]. In contrast, daily stresses in people’s lives may be far more insidious in their long-term biological consequences. A previous study utilising the Finnish Twin Cohort showed a heightened association between accumulation of stressful life events, in particular of those of greatest a-priori-interest, such as divorce or death of a loved one, and increased risk of breast cancer incidence during 15 years of follow-up [11]. Our current study extends the follow-up period to 36 years after baseline assessment (1981) and includes a greater number of incident breast cancer cases. Life events occurring in the five years prior to assessment are qualified as “stressful” and quantified based on the 21-question list described by Lillberg et al. (2003) [11], with questions ranging from more commonplace stressors, such as acquiring a loan, to the more unexpected and traumatic end of the spectrum, such as a death of a loved one. The increase in statistical power, attributed to the longer follow-up time and, consequently, the number of breast cancer cases, reasserts previous findings and shows that accumulating multiple stressful life events, regardless of their type, significantly increase the risk of breast cancer incidence in the long-term. Leveraging twin pair analyses, we established that the heightened risk of breast cancer is attributable to factors that are not shared between the twins, i.e., the increased risk in breast cancer incidence following stressful life events is independent of genetic background. Prior research has indicated that associations of stressful life events with breast cancer risk are not consistently observed. There are limitations in these analyses, such as potential recall bias in retrospective studies and the challenge of drawing temporal conclusions from studies with varying follow-up periods, as shorter follow-up periods could limit the ability to capture the full effect of stressful life events on breast cancer development, which can take years to manifest.

A noteworthy feature of our dataset is that individuals were born across different decades, often prior to other studies referenced in the literature [11]. While no clear trend in genetic risk across these cohorts was visible when stratifying the data based on decade of birth, we did observe a higher HR for stressful life events in individuals born before 1950. The observed difference in the risk ratio associated with stressful life events may be reflective of the turbulent times in Finnish history, such as the 1930s depression, World War II and the immediate post-war period, when Finland was much less developed than it is currently. Only from the 1950s did economic and social development happen within a long period of peace. Another notable finding is that of oral contraceptive use, whereby women born prior to 1950 had 28% increase in hazard ratio for breast cancer as compared to virtually no effect among those born after 1950. This may be due to the changing doses of oral contraceptives, as pills transitioned from high-dose oestrogen formulations (50-100 mcg) in the 1960s to lower doses (10-20mcg).

The epidemiology of breast cancer in Finland is evolving. The cumulative hazard for those born after 1950 in the current cohort was shown to be significantly higher than those born earlier (Supplemental Figure 1), reflecting the general trend of increasing breast cancer incidence in the Finnish population (Finnish Cancer Registry Incidence Figure from www.cancer.fi). However, birth cohort analyses in the entire Finnish population based on the NordCan database suggest that the age-specific breast cancer incidence of women born from 1950 onwards is no longer increasing (Supplemental Figure 2, Nordcan visualisation from https://nordcan.iarc.fr/). The same change can be seen for Sweden as well in the Nordcan data. Our stratified analysis delineates the importance of timeframe when considering risk associations with breast cancer. There is potentially a paradigm shift in the epidemiology of breast cancer indexed by a major change in incidence for women born in the post-WWII decade in at least two Nordic countries. Our suggestive findings for stressful life events point to greater resilience among women born in the 1950s. These intriguing but preliminary findings warrant further investigation to consolidate the reasons underlying these changes in the epidemiology of breast cancer.

We used genetic and epigenetic data to explore potential explanations for the association of stressful life events and incident breast cancer. Firstly, genetic breast cancer risk assessment revealed, at most, a weak association between the breast cancer PRS and breast cancer incidence in our cohort. The hazard ratios ranged between 1.01 and 1.23, which are smaller than those reported in previous study by Mars et al. (2020) (HR = 1.71; 1.68–1.75) [4]. However, the effect size of the used breast cancer PRS was comparable in other European populations [24]. As our cohort comes from the same population of Finns as in Mars et al. (2020) [4], the smaller effect size is probably an indicator of other differences between the FinnGen cohort used by Mars et al. (2020) [4] and our twin cohort. The potentially diluted observation between breast cancer PRS and breast cancer incidence may be due to the substantially longer follow-up period compared to previous studies, as environmental risk factors may exert a stronger influence over time, masking the genetic contribution. While our findings on breast cancer PRS showed only weak association with breast cancer incidence, it is important to acknowledge limitations of PRS itself. The current PRS model explains a relatively small proportion (less than 1%) of the total breast cancer risk [4]. Therefore, it may not fully capture the estimated 31% risk attributable to genetic factors based on twin studies [2].

We did not see an association between breast cancer PRS and stressful life events score, suggesting that the effect of life events on breast cancer risk is not mediated by shared genetic liability. Though life events are considered to be external to the persons affected by them, the response to such events is in part affected by biological factors and hence possibly by genetic effects [25–27]. Likewise, the subjective experience of stressful life events is variable, and thus reporting of them may vary.

In contrast to the genetic findings, our epigenetic results demonstrate an association between exposure to life events and specific DNAm patterns. These same DNAm patterns are also linked to an increased risk of breast cancer, particularly to breast cancer risk coming from environmental exposures (Bode et al, 2024) [17]. However, to ensure that this association with stressful life events is not simply a consequence of an underlying undiagnosed cancer, we have shown in a separate dataset of healthy twin pairs that similar effects of stressful life events on the environmental breast cancer risk DNAm patterns are also present there. These results suggest that the association between stressful life events and DNAm is likely independent of a subsequent breast or other cancer diagnosis. Prior research supports this connection: studies have shown that stress exposure can modify DNAm patterns [28–32]. Additionally, stressful life events can disrupt a woman’s endocrine hormone balances [18]. Changes in endocrine hormones, in turn, are known to influence DNAm [33]. This suggests that life events might leave epigenetic marks on DNA directly or indirectly via changes in endocrine hormone levels, which could potentially affect the function of breast cancer susceptibility genes and thereby alter biological pathways that promote breast cancer development. The observed DNAm changes appearing, on average, 16 years after life event exposure but up to 11 years before breast cancer diagnosis strengthens the potential link between life events, and breast cancer development via DNAm, as these DNAm patterns are present during a critical time window for breast cancer initiation. However, future research is crucial to establish a causal relationship between DNAm and breast cancer risk, as well as with other cancers, especially ones sharing similar risk factor profiles or same driver genes. This could involve investigating the functional consequences of these DNAm changes on relevant pathways. Additionally, studies comparing DNAm in breast tissue instead of blood samples would provide valuable insights.

Our study has some limitations: Our results are confined to women of European origin, specifically of Finnish descent, underscoring the necessity of conducting studies on diverse populations before making broader generalisations. We acknowledge that some of our analyses may lack statistical power due to limited sample size. Our study did not include data on high-impact breast cancer variants, such as those in BRCA1 and BRCA2, limiting a comprehensive examination of their potential influence on breast cancer risk in the current study as these variants do interact with the PRS in a non-additive manner, as demonstrated by Mars et al. (2020) [4]. However, given the number of breast cancer cases, any analyses of high-impact breast cancer genes would be severely constrained.

CONCLUSION

This investigation contributes to our understanding of the delicate aetiology of breast cancer and further solidifies a concerning trend: stressful life events appear to be a major long-acting risk factor for breast cancer. Cumulative stressful life events emerged as a potent risk factor, with a hazard ratio exceeding that of other measured factors across generations, especially for women born prior to 1950. While genetics has an effect, it is likely weak, as our analysis suggests that the majority of breast cancer incidence is likely influenced by non-genetic factors. Additionally, epigenetic changes linked to breast cancer risk were associated with stress exposure.

Remarkably, these changes persisted on average at least 16 years after the stressful life events and, on average, 11 years before diagnosis. Our findings highlight the critical need to understand the role of stress in breast cancer aetiology.

Declarations