Findings from this study suggest that genetic variation in DNA repair and oxidative stress pathways play a role in odds of pain development prior to and after 6 months of continuous AI therapy and pain severity prior to AI therapy in women with hormone receptor positive early-stage breast cancer, but different polymorphisms in different genes are significant at different timepoints. This finding could, in part, be due to the pain type experienced by the women at each timepoint. Anatomical location of pain was different between the women with breast cancer prescribed AI therapy and healthy controls at both timepoints. These differences may represent physiologically different mechanisms of pain. At baseline, most pain reported by women with breast cancer was in the breast/axilla area of the body. This finding is supported by studies suggesting that nearly 25% of women with breast cancer experience substantial levels of breast pain in the first six months following breast cancer surgery [47] and up to 50% of these women may develop chronic breast pain [49]. The area around the breast is the most frequently reported site of pain post-breast cancer surgery, followed by the axilla, arm, and the side of the body [46, 49]. At 6 months, pain was reported by women with breast cancer in more distal areas. This is in concordance with previous research findings that symptoms of AI-induced musculoskeletal pain emerge at around 2 to 3 months and are the most severe at 6 months [48, 50].
Although breast/axilla pain was the most frequently reported pain location at baseline, there still were many women with distal pain prior to initiation of AIs. Because pain data was subjective and no objective measures were used, it is difficult to distinguish pain type experienced at each timepoint. Based on the literature, we concluded that breast/axilla pain was attributed to breast cancer surgery as this pain was experienced by nearly 25% of the women with breast cancer [46, 49, 51]. On the other hand, it was difficult to discern the source of reported distal pain, as it could be AI-induced musculoskeletal pain, but it could also be more general pain experienced by postmenopausal women because of estrogen loss [48, 52].
The odds of having pain versus no pain at baseline were contingent on participants having one or more minor alleles from a combination of four SNPs from two oxidative stress genes (Table 3). This is interesting considering that it has been suggested that decreased estrogen is a risk factor for oxidative stress because estrogen is involved in the expression of antioxidant genes [53]. There were no DNA repair genes included in the GRS calculation, as none of the DNA repair genes were independently significant in the relationship between the individual SNP and average pain score. A possible reason for this could be that oxidative stress leads to DNA damage and the need for DNA repair [54]. Therefore, it could be that we see SNPs in oxidative stress genes having a greater impact on whether someone has pain in earlier stages, and SNPs in DNA repair at later timepoints when DNA repair enzymes are more active in repairing damage. Also, breast cancer cells can be susceptible to oxidative damage and have high levels of oxidative stress that may vary depending on the presence and progression of the breast cancer tumor and the different cancer treatments, including chemotherapy and surgery [10].
When considering the significant SNPs in this study, the effect, main (associated with the odds of pain and/or pain severity regardless of breast cancer or treatment type) or interaction, in this case the interaction of treatment (AI ± chemotherapy) compared to healthy controls, or both, is important. Understanding these associations helps us to understand the association of each SNP to the GRS calculation. For instance, in the baseline GRS calculation on the odds of having pain, the CAT-rs566979 SNP is a main effect, so regardless of breast cancer status or prescribed treatment/disease severity, this particular SNP influences pain. On the other hand, in the same calculation, the SEPP1-rs230819 polymorphism is a statistically significant interaction effect for the prescribed treatment of AI alone; this means that the influence of this SNP depends on cancer status and prescribed treatment/disease severity. Lastly, looking at the baseline GRS calculation for pain severity in women who report pain, we see that ERCC5-rs4150355 has both a main and an interaction effect, meaning that this SNP influences the odds of pain presence regardless of cancer status and prescribed treatment/disease severity, but also that the significance depends on prescribed treatment/disease severity, which in this case is AI + chemotherapy.
In this study, we found that different alleles in DNA repair and oxidative stress genes have different impacts on pain at different timepoints. We did not see any significant relationships between polymorphisms and perceived average pain at later timepoints (12 and 18 months). This could be due to a variety of reasons [55]. The most likely reason is that the sample sizes were too small to detect any differences. As the parent study was designed to explore differences in cognitive function, pain was not an original main outcome; therefore, women who stopped taking AI therapy were dropped from the study and no further data were obtained from these women.
There were differences in pain severity and location between women with breast cancer and healthy controls (Table 2, Online Resource 5). Not only did women with breast cancer report significantly higher severity, but they also had a more distinct location pattern than healthy controls. Breast cancer tumors are heterogenous. To account for this difference in disease characteristic, participants with breast cancer were classified by their prescribed treatment group. In this study, there was no difference in pain severity or location between the women who received AI + chemotherapy and women who received AI alone. This finding could be due to the fact that all women with breast cancer underwent breast cancer surgery, and the perceived pain may be more related to the surgery and not the biology of breast cancer. Breast/axilla pain was only reported by women with breast cancer, while healthy controls reported no pain or more general distal pain.
Results of this study suggest that the variation in polymorphisms in DNA repair and oxidative stress genes are important to the occurrence and severity of perceived average pain. However, the impact of these polymorphisms appears to be different depending on study timepoint. This can be visualized in Online Resources 1 and 2, where the variability is smaller at earlier timepoints, but the distribution is such that the GRS does significantly impact the odds of having pain (Online Resource 1) and severity of pain (Online Resource 2). Yet, at later timepoints, the variability is too large to detect significance. Further research with a larger sample is needed for these timepoints.
Limitations and Future Directions
In this exploratory analysis, we focused on polymorphisms that had previously been used in a study designed to examine cognitive function and therefore we may not have evaluated variability in all genes in the DNA repair and oxidative stress pathways important in variability of average perceived pain. The small sample size limited the ability to use genotypes to conduct the genetic analyses, so instead we used the possession of one or more major alleles, limiting our ability to determine the gene-dose effects. Moreover, our sample consisted of postmenopausal women with early-stage hormone receptor-positive breast cancer who reported being primarily White. Therefore, the ability to generalize the results to non-White, premenopausal, hormone receptor-negative, and more advanced cancer patients, is limited. Also, because the parent study was designed to examine cognitive function in women prescribed AI therapy, data were no longer collected for those participants who stopped taking AI therapy, limiting the ability to follow these women beyond their withdrawal. Future studies should explore the polymorphisms in a larger, more diverse sample that uses genotypes rather than the major allele to determine the effect of variability in these polymorphisms on average perceived pain.