Despite the overall higher incidence of breast cancer in European-American (EA) women as compared to African-American (AA) women, AA women are more likely to be diagnosed with breast cancer before age 40 and to have tumors with aggressive pathological characteristics, including high tumor grade, lack of expression of estrogen and progesterone receptors (ER, PR) and HER2 (triple-negative breast cancers), and additional features of basal-like breast cancer (ER−/PR−/HER2−/cy5/6+/EGFR+) [1]. Triple-negative and basal-like breast cancers are recognized to be associated with a considerably poorer prognosis than other breast cancer subtypes [2]. Although the role of genetic factors, including polymorphisms in the BRCA1, BRCA2, ATM, CHEK2, p53, PTEN, NBS1, RAD50, BRIP1, and PALB2 genes have been extensively studied in breast cancer in EA populations [3], there are few studies of the role of common variants in nuclear gene(s) related to mitochondrial function in the etiology of breast cancer in AA women.

A recent study demonstrated that tumor-cell mitochondrial DNA copy (mtDNA) number correlates with tumor progression as well as patient prognosis and disease-free survival in breast cancer [4]. Cellular mtDNA copy number (or content) is controlled by the nuclear-encoded POLG gene encoding the only known mtDNA polymerase (polymerase-gamma) in humans. Our previous research suggests that mutations in the POLG gene may result in depletion of mtDNA and confer breast cancer phenotype [5]. Human POLG consists of an exonuclease domain with three exonuclease motifs, I, II and III, and a polymerase domain with three polymerase motifs, A, B and C, along with an intervening linker region [6]. The exonuclease domain of POLB, which is responsible for the proof-reading activity of the encoded enzyme, harbors a CAG repeat region in exon 2. The contraction of CAG repeats in POLG affects its expression [7], and an expanded CAG repeat sequence seems to confer toxic functions on the protein through protein-protein interactions [8]. The CAG repeat polymorphism has been studied in several diseases, including male infertility and neurodegeneration [9]–[11]. An association of the POLG-CAG repeat expansion with testicular cancer has also been reported [12]. Previous studies have also suggested that variation in the number of CAG repeats in androgen- and estrogen-receptor genes influence the risk of breast cancer [13]. However, the significance of the CAG repeat polymorphism of POLG in breast cancer has not been investigated. The purpose of this study was to evaluate the association of POLG CAG repeat length with breast cancer risk in the Women's Circle of Health Study (WCHS), an epidemiological case-control study of breast cancer in AA and EA women.

The CAG fragment analysis was done in a blinded manner and repeat length was carefully assigned based on the peak size. After the initial observation, we eliminated 12 samples because they showed poor signal due to inadequate DNA. In this subset of the WCHS, there were no statistically significant differences between cases and controls with respect to age at menarche, age at first pregnancy, number of children, menopausal status, and family history of breast cancer (Table 1). We identified 7 breast cancer cases (7%) and 6 controls (6%) in which the common 10/10 POLG-CAG repeat was absent. The common 10/10 CAG repeat occurred in 52 cases (55%) and 56 controls (60%); heterozygous 10/non-10 CAG repeats were observed in 36 cases (38%) and 31 controls (33%) (Table 2). The representative CAG repeat contraction and expansion variants are shown as electropherograms (Figure 1).

Compared with 10/10 repeats, the crude odds ratios (ORs) for 10/11, 10/12 repeats, and all other repeats were 1.16 (95% CI, 0.58–2.32), 0.93 (95% CI, 0.36–2.35), and 1.31 (95% CI, 0.46–3.76), respectively (Table 1). After adjustment for age, age at menarche, age at first full-term pregnancy, menopausal status, first-degree family history of breast cancer, and BMI, the adjusted ORs for 10/11 and 10/12 repeats were 1.15 (95% CI, 0.50–2.63) and 1.25 (95% CI, 0.45–3.46), respectively. In the case of CAG repeats other than 10/11 or 10/12, point estimates suggested an increased risk of breast cancer, but no statistically significant difference was observed in the frequency of CAG repeats between cases and controls (OR 1.74; 95% CI, 0.49–6.21, Table 2).

Mitochondrial POLG is the only polymerase known to be involved in human mitochondrial DNA replication. POLG contains CAG trinucleotide repeats in the coding region, and CAG repeat sequences are described to be highly unstable, leading to expansion or contraction of the repeat sequences [8]. In a recent study of the POLG gene, we reported the CAG repeat expansion occurs in 20% of breast cancer patients [5]. This pilot study suggests the possibility of an association between altered POLG-CAG repeat-length and an increased risk of breast cancer in AA women.

The presence of expanded or contracted CAG repeats has been linked with several diseases, including the human polyglutamine diseases [8]. Other epidemiologic studies have shown either a positive or negative correlation of CAG repeats with sporadic breast cancer, and one study reported an inverse correlation between POLG-CAG repeat length (r = −0.81) and the age of onset of disease in Friedreich's Ataxia (FRDA) patients [13]. In another study dealing with myotonic dystrophy, no significant difference in the frequency of CAG repeats was seen between cases and controls [7]. Studies of CAG-repeat polymorphisms in antiretroviral therapy-associated peripheral neuropathy similarly showed no significant associations [15]. However, CAG repeat-length polymorphisms have been associated with male infertility; the common 10-repeat variant of the CAG repeat was found to be significantly more prevalent among men with oligospermia [9] and unexplained subfertility [16].

There are two additional reports associating POLG-CAG repeat variants with cancer risk. Interestingly, a study of CAG repeat polymorphisms in the POLG gene in testicular cancer showed 36 (74%) wild-type homozygotes and 13 (26%) lacked one or both wild-type alleles, with the 10/11 variant in 10 patients and the 10/12, 10/6 and 11/11 variants in one patient each, suggesting that variants of the DNA POLG1 gene were more frequent in testicular cancer patients than in healthy men [12]. Another study suggested that the CAG polymorphism in POLG may be a contributing factor in the pathogenesis of testicular seminoma [17]. It should be emphasized, however, that existing studies in of POLG-CAG repeat variants in cancer compared the frequency of these variants in tumor tissue to the frequency in either surrounding normal tissue or blood samples. In contrast, our study evaluated germline polymorphisms in POLG in both breast cancer cases and in controls. Results of studies may differ depending on the DNA source due to additional clonal mutations that can occur in tumor cells.

This study is the first to investigate the role of CAG repeat polymorphisms in the nuclear POLG gene, which encodes mitochondrial DNA polymerase-gamma, in breast cancer risk. While our findings suggest a possible correlation between CAG repeat length and risk of breast cancer in AA women, the size of this genetic study sample was small, as evidenced by reduced power to demonstrate significant associations of known breast cancer risk factors such as age at menarche and family history of breast cancer with cancer risk in this study. Larger, preferably population-based, studies are needed in order to draw any firm conclusions about the role of POLG-CAG variants in breast cancer risk in AA or other populations.