Modification of the association between recreational physical activity and survival after breast cancer by promoter methylation in breast cancer-related genes

With an estimated 40,000 deaths in 2017, breast cancer is the second leading cause of cancer-related death in the United States [1]. Women who engage in physical activity prior to the diagnosis of breast cancer have better overall survival than those who do not [2], but the mechanisms of this association are unknown. Given that only 20% of the U.S. population achieves the Centers for Disease Control and Prevention’s physical activity guidelines [3], improved understanding of how physical activity influences breast cancer prognosis could have significant public health impact.

Epigenetics is the study of functionally relevant changes to the genome that do not involve a change in the nucleotide sequence. DNA methylation is the most extensively studied epigenetic modification and involves the addition or removal of methyl (-CH₃) groups at CpG dinucleotides that influence gene regulation [4]. DNA methylation can be measured in a range of tissues, including tumor and blood [5], and has been associated with breast cancer prognosis in several studies, including our own [6, 7]. Although methylation signatures are largely established during embryogenesis [8], DNA methylation (and other features of the epigenome) may be modified throughout the life course as a result of both behavioral and environmental stimuli [9], including physical activity [10]. Interactions between the environment and DNA methylation may, therefore, inform prognostic outcomes among women diagnosed with breast cancer.

In a population-based sample of women diagnosed with first primary breast cancer, we aimed to understand whether the association between prediagnostic recreational physical activity (RPA) and all-cause or breast cancer-specific mortality was modified by gene promoter methylation (which regulates gene expression) in a panel of 13 breast cancer-related genes (APC, BRCA1, CCND2, CDH1, DAPK1, ESR1, GSTP1, HIN1, CDKN2A, PGR, RARβ, RASSF1A, and TWIST1) measured in tumor tissue. Similarly, we sought to determine whether the RPA-mortality association was modified by global DNA methylation (a marker of genome stability) using two methods to assess white blood cell methylation: long interspersed nucleotide element 1 (LINE-1), which approximates levels in repetitive elements [11], and the luminometric methylation assay (LUMA), which estimates methylation at CCGG sites [12]. We hypothesized that methylation of oncogenes (or lack of methylation in tumor suppressor genes) and high prediagnostic physical activity engagement would result in lower all-cause and breast cancer-specific mortality among women diagnosed with first primary breast cancer. We also hypothesized that physical activity and low LUMA (high LINE-1) would work in synergy to reduce mortality following a breast cancer diagnosis.

For this project, we used resources from the follow-up component of the Long Island Breast Cancer Study Project (LIBCSP), a population-based study. Details of the study design and participants for this component have been described previously [13, 14].

In this population-based follow-up study of 1254 women diagnosed with first primary breast cancer, we found that the overall improved survival among women with any lifetime prediagnostic RPA appeared to be modified by gene-specific methylation profiles. We observed substantially improved survival with high lifetime prediagnostic RPA in women with a tumor-methylated APC, CCND2, HIN1, or TWIST1 gene promoter compared with active women with unmethylated gene promoters. A more pronounced risk reduction was observed for breast cancer-specific mortality for the interaction with HIN1; however, we were unable to evaluate interactions with breast cancer-specific mortality owing to small numbers. We found no interaction between RPA and global methylation as measured by LINE-1 and LUMA. Our findings suggest that the inverse association between RPA and mortality after breast cancer may depend upon gene-specific methylation profiles.

Improved survival with RPA among women with breast cancer has been observed in many epidemiologic studies [2], including our own [14]. Also, we [6] and others have reported associations between gene-specific methylation and prognosis [34]. However, to our knowledge, no previous investigation has considered gene methylation as a potential modifier of the RPA-mortality association, despite strong biologic plausibility. Not only does physical activity reduce adiposity and its numerous metabolic correlates, but it is itself thought to reduce markers of inflammation, alter immune functioning, and lower circulating insulin [35, 36]. These pathways have been linked to aberrant DNA methylation, altering several genes implicated in breast carcinogenesis [37‐40]. Collectively, these data suggest that the mechanisms underlying the inverse association between RPA and mortality may be facilitated and/or altered by inflammation-related methylation changes.

In the present study, we found that the improved survival after breast cancer with high RPA was greatest among patients with methylated APC, CCND2, HIN1, and TWIST1 promoters. APC and HIN1 are candidate tumor suppressors thought to be involved in breast carcinogenesis [41, 42]. Our observation of decreased mortality among very active women with APC or HIN1 methylation is counter to our a priori hypothesis of lowered risk of death among active women with unmethylated (active) tumor suppressor genes, although we did observe that in the highest RPA group, there was no statistical difference in the effect among APC methylated and unmethylated cases. We observed risk reductions at both low and high RPA engagement among women with methylated HIN1 promoters. Methylation of HIN1 is linked to gene silencing, reduced expression, and loss of apoptosis [43]. HIN1 is a putative growth inhibitory cytokine thought to be inactivated at the earliest stages of breast tumorigenesis and silenced in the majority of sporadic breast carcinomas [44]. This may suggest that prediagnostic RPA could help overcome the deleterious effects of HIN1 inactivation in breast carcinogenesis, thereby improving survival outcomes.

The exact roles of CCND2 and TWIST1 in breast cancer are unresolved. CCND2 is important in cell cycle regulation and has been cited as both a tumor suppressor gene [45] and an oncogene [46]. Inactivation of CCND2 is thought to occur via promoter hypermethylation, which may be an early, though infrequent (about 11%), event in malignant breast cancer transformation [45, 47, 48]. In our study, we found lower all-cause mortality among women with promoter methylation (or loss of CCND2 expression) in tumor tissue. This may reflect synergy between physical exercise and inactivation of the CCND2 oncogene, particularly among women with low RPA. TWIST1 is an antiapoptotic and prometastatic transcription factor, and methylation of the gene promoter has been observed frequently in malignant breast tissue [49]. We observed pronounced reductions in all-cause mortality among physically active patients with TWIST1 methylation, which is consistent with our a priori hypothesis of synergy between the presumptive oncogene and RPA.

Although our population-based study of women with breast cancer was carefully conducted and included comprehensive exposure assessment and a long follow-up time, several potential limitations should be considered. First, information on RPA was collected systematically by trained interviewers [13]; nonetheless, there is potential for nondifferential measurement error, which would result in reduced effect estimates. However, LIBCSP investigators used a comprehensive, open-ended approach to query women on their lifetime RPA habits. This approach has been shown to elucidate important relationships between RPA and breast cancer in the LIBCSP [14, 16] and is consistent with other findings [50]. Second, postdiagnostic RPA, which likely influences prognosis [51], was not considered in this investigation, owing to small cells after stratification by methylation status. Third, we were limited to a panel of 13 biologically relevant genes and 2 global assays. However, studies employing more robust panels that interrogate hundreds of thousands of CpG sites are at high risk for false-positives, generally lack biologically driven hypotheses, and perform modestly using archived tumor samples [52, 53]. Additionally, we were limited by the use of conventional MSP PCR assays for three of the genes. However, where quantitative MSP PCR assays have the advantage of providing a quantitative estimate of methylation, the conventional MSP assay is a highly sensitive method to classify individuals by methylation status, which mitigates the threat of biased results [54, 55]. Finally, the racial homogeneity of our study population restricted our ability to explore potential variation by intrinsic subtype or by race, both of which are known to associate with prognostic outcomes [56]. Nonetheless, the largest hormonal subtype of breast cancer diagnosed among U.S. women of any race is ER⁺PR⁺ [57], which continues to increase with time [58] and is the predominant subtype of breast cancer diagnosed among LIBCSP study participants. When we restricted our findings to women with only hormone-responsive breast tumors, results were similar to those for all women.

To our knowledge, we are the first to show, using resources from a population-based follow-up study, that promoter methylation of APC, CCND2, HIN1, and TWIST1 may modify the inverse association between prediagnostic RPA and all-cause mortality following a breast cancer diagnosis. With the exception of HIN1, which was suggestive of breast cancer-specific mortality, power was limited for examining potential modification of the association between RPA and breast cancer-specific mortality. Although our results require confirmation in cohort studies with a larger number of women with breast cancer and more comprehensive gene coverage, they suggest that DNA methylation may play an important role in associations between physical activity and improved survival among women with breast cancer.