Breast pain in a diverse population of breast cancer patients

Background Breast cancer is the most common cancer in women. The majority of women with breast cancer present with early stage disease requiring surgical management. Post-operative breast pain has been reported to be anywhere from 25–60%. Racial disparities in cancer treatment-related symptom burden are known and linked to worse treatment outcomes. There is sparse data regarding racial/ethnic differences in breast pain among breast cancer patients. We evaluated the prevalence of breast pain in breast cancer patients and characterized the pain using a modied short-form McGill pain questionnaire in our diverse population. and and a medium effect size (Cohen’s f equal to 0.25 or 6% of the variance explained), 53 patients per group provide 80% power. For the logistic regression analyses, assuming a Type I error rate (alpha) of 0.05, and an odds ratio of 3.35, 50 patients per group provide 80% power. The SAS 9.4 (SAS Institute Inc., Cary, NC) software performed all analyses. The LOGISTIC procedure ran logistic regression analyses on the binary (yes or no) pain variable. The GENMOD procedure ran linear regression analyses on the continuous pain variable. The predictor variables in both models were age, ethnicity,, surgery, radiation, chemotherapy, endocrine therapy, lymph nodes, and location of questionnaire. We ran both univariable (unadjusted) and multivariable (adjusted for all other variables) models. We did not run analyses by subgroups or interactions. We addressed missing data using listwise deletion. We ran analytic methods for a cross-sectional study with simple sampling. a recent review, the variation in disease, treatment mode and symptom progression in various studies make it dicult to draw conclusions about chronic pain from both the prospective and retrospective literature . The literature has focused predominantly on risk factors for the development of chronic pain, however some studies have evaluated potential chronic pain mechanisms which may be an important key to determine specic treatments. Studies have evaluated genetics variations in cytokine and potassium channel genes to identify persistent breast pain phenotypes 38,39 and their potential mechanisms 40 . These are interesting ndings, however their translational merit has yet to be realized.


Background
Breast cancer is the most common cancer in women 1 . It has been predicted that the worldwide incidence of female breast cancer will approach 3.2 million per year by 2050 1 . Prevalence rates for persistent pain following breast cancer surgery are reported for up to 60% of patients [2][3][4][5] . Approximately 30% of women report pain in the affected breast pre-operatively 6 . Not surprisingly, pre-operative breast pain is a predictor of post-operative breast pain 2,6−8 . Younger age, use of radiotherapy, more invasive surgery, and acute post-operative pain have been identi ed as predictors of persistent pain after breast cancer surgery 2,4,6,8−10 .
Given the large number of women who will undergo breast cancer surgery, it is important to address persistent breast pain.
Racial disparities in cancer treatment-related symptom burden are known and linked to worse treatment outcomes 11 . Minority women describe more symptom intensity and distress with breast cancer treatment 3,12,13 . There are also reported differences in physical, social and emotional well-being among cancer survivors of different racial/ethnic groups 14 . There is limited data regarding racial differences in patients' treatment-related symptom management experiences. Some studies have focused on particular racial/ethnic groups. Within Black and Latina breast cancer patients, differences have been found in management of symptoms 15 and provider communication 16 . Routine symptom assessment for all breast cancer survivors, as well as clinicians' management of symptoms for racially diverse cancer patients, need to be more thoroughly studied and addressed.
Several studies have looked at factors predicting post-operative and persistent breast pain, but to our knowledge none of these studies have reported on differences among multiple ethnic groups 17 . A study performed in China reported 28.5% of women reported persistent pain after surgery with 50.5% of the people reporting persistent pain, reporting sensory disturbance and 4.2% reporting phantom breast pain 18 . The ethnic population of Hawaii is diverse. Hawaii has no racial/ethnic majority group, based on population size. About half of the population is of Asian heritage (Japanese, Filipino, Korean, Chinese), about a quarter is of European ancestry (White) and about 20% Native Hawaiian. Native Hawaiians have the highest rates of newly diagnosed cases and deaths from breast cancer compared to White, Japanese, Filipino and Chinese patients 19 .
The aim of this study was to determine the prevalence of breast pain, its associated factors and to characterize breast pain in a diverse population of breast cancer patients. Further elucidating the characterization of breast pain in breast cancer survivors has translational merit to develop interventions that are relevant to all backgrounds.

Study Population
We conducted a cross-sectional survey of breast cancer patients from the Hawaii Cancer Consortium over a 6-week period between 2019 and 2020. To ensure a broad selection, surveys were offered in 4 clinical areas: outpatient oncology clinics, breast cancer survivorship groups, social work in person visits, and as an online link. Patients were eligible for the study and were given the opportunity to participate if they 1) had a diagnosis of breast cancer 2) were Englishspeaking, and 3) had not completed the survey at a prior time.

Study Procedures
The study received approval through the University of Hawaii and local hospital Institutional Review Boards. Eligible patients were identi ed by trained clinical staff and were offered the questionnaire if they had a diagnosis of breast cancer. Interested patients were given the paper survey while they waited for their oncology appointment or at the end of the survivorship group meeting or social work appointment. Completed surveys were returned directly to front-desk staff and were picked up weekly by the research team. Fliers were also provided and given to patients if they were interested in the questionnaire but were not able to ll out the information at that time. The ier contained information about the purpose of the assessment and included a link to an online version of the questionnaire. All anonymized survey responses were entered into OnCore or downloaded and were double-entered for accuracy.

Assessment
The rst page of the assessment materials included an introductory page describing the study goals, the anonymous nature of responses, and the estimated time for completion. If the participant reported breast pain in the last 6 months they were asked to complete all items of the questionnaire. If they did not report any breast pain in the last 6 months, they were asked to skip all pain related questions which included 15 items. Completing the entire assessment took between 5-10 minutes for a majority of the participants.
The online survey was administered using Qualtrics-a survey management software supported by the Research Systems o ce at the University of Hawaii Cancer Center. The study team designed the content of the questionnaires and provided template les to the technical team. All questions provided were tabulated into the survey designer user interface and aesthetically formatted. Validation logic and skip logic were added to ensure responses were complete and user experience was not impeded. The survey was published and associated with an internet address. This address was embedded into a link that was shared with study participants. Survey response data was monitored for discrepancy patterns via the data analysis dashboards provided by Qualtrics. This interface was also used to perform data exports once the data was locked. Permissions set by the technical team ensured that only assigned users were able to edit forms and no users were given permissions to change subject response data.
Measures Demographic variables included age and self-reported ethnicity, whereas cancer status included various indicators of the current diagnosis and treatment status (breast cancer diagnosis, type of surgery-lumpectomy or mastectomy, number of lymph nodes removed, use of radiation, chemotherapy and/or endocrine treatment). The breast pain questionnaire was based on the short-form McGill pain questionnaire 20  Scale. Breast pain was assessed using multiple variables, including quality, intensity, amount, pattern, duration, timing, location and associated aggravating or alleviating factors. In addition, if pre-menopausal, patients were asked if there was any association with menstrual cycle and all patients were asked if breast pain affected work, sleep or sexual activity. Participants were asked to also discuss any other pain either associated with breast pain or not and if any medications or other modalities were used to relieve breast pain. Statistical Methods Within the funding and time constraints, we collected the maximum sample size possible. We required that this exceeded the minimum for su cient power. Survey responses and data comparison were summarized as frequencies and percentages for categorical variables and as medians for continuous variables.
We used standard regression methods for continuous outcomes and logistic regression for categorical outcomes. In multivariable analyses, we included all study variables as covariates to adjust for confounding. We chose the simplest relevant groupings with su cient observations. For the linear regression analyses, assuming a Type I error rate (alpha) of 0.05, and a medium effect size (Cohen's f equal to 0.25 or 6% of the variance explained), 53 patients per group provide 80% power. For the logistic regression analyses, assuming a Type I error rate (alpha) of 0.05, and an odds ratio of 3.35, 50 patients per group provide 80% power. The SAS 9.4 (SAS Institute Inc., Cary, NC) software performed all analyses. The LOGISTIC procedure ran logistic regression analyses on the binary (yes or no) pain variable. The GENMOD procedure ran linear regression analyses on the continuous pain variable. The predictor variables in both models were age, ethnicity,, surgery, radiation, chemotherapy, endocrine therapy, lymph nodes, and location of questionnaire.

Demographics
The median age of the respondents with breast pain was 58.5 years and 63 years for patients without breast pain (range, 29-98y; see Table 2). There were no statistical differences seen in breast pain likelihood according to racial/ethnic group. There were not any significant findings on logistical analysis on the other variables included in the analysis.
However, on multivariate regression analysis, we did find significant racial/ethnic differences in the amount of breast pain, where therapy had a statistically significant increase in breast pain (2.4 vs 1.83, p=0.002). There were no differences in breast pain according to the type of breast or axillary surgery. There were statistically significant differences in breast pain seen depending on survey location.
Relative to online assessment, paper assessments in clinic offices reported less breast pain (Castle 1.
For mild pain, the most common descriptor was aching (n=20), for moderate pain the most common descriptor was sharp (n=9) and was a mix for severe pain--heavy, tender, shooting and throbbing (n=2 respectively) without a most common descriptor (Fig. 3).

Summary of Findings
In this cross sectional study of breast cancer patients at the Hawaii Cancer Consortium, it was found that thirty-ve percent of participants reported breast pain. Although there were no statistically signi cant differences seen in breast pain according to racial/ethnic group, there were signi cant racial/ethnic differences in the amount of breast pain reported. Non-white participants reported less breast pain compared to White participants in our study. Participants who were of younger age reported increased breast pain for most age groups. Breast pain differences were seen according to some treatments-those who received chemotherapy and radiation were experiencing less pain and those undergoing endocrine therapy were experiencing more pain-however differences were not seen for the type of surgical management. Online and survivorship group assessments reported more breast pain compared to clinical o ce assessments. Sensory qualities of pain were more commonly reported than affective characteristics. Although overall breast pain was mild and intermittent, it affected participants' sleep, work and sexual activity. Breast pain treatment modalities reported were varied and a substantial percentage of participants reporting pain also report that they have not identi ed anything to alleviate the pain. This indicates an opportunity to identify intervention for breast pain.
Participants completing the questionnaire online reported more breast pain compared to paper surveys. Although the numbers are small, a majority of participants from the survivorship groups and those who completed the survey online had breast pain, suggesting that the location from where the information is being collected is important. Dedicating attention to this common symptom in a variety of formats (online and paper) and locations (clinic and survivorship groups), has allowed more patients to describe their breast pain. Patients may not want to discuss breast pain with their clinician during an o ce visit that is often focused on evaluating for recurrence or other important aspects of cancer care, when the breast pain symptoms are mild and intermittent in nature, potentially suggesting that the report of pain online might be more accurate. Despite reports of breast pain affecting their sleep, work and sexual activity, obtaining breast pain information during a routine visit may not be seen as a priority for either the clinician or patient. Whereas in a survivorship group setting, discussions are commonly center around quality of life issues, suggesting that part of why people are in the group is because they want to share these concerns 22 .
Our study found that 35.4% of participants had breast pain, which is similar to previous studies reporting a range of 20-60% of breast cancer survivors reporting persistent pain [2][3][4][5]23 . The existing data arises from a predominantly White demographic, where our cohort includes Asian, Native Hawaiian and mixed heritage participants. In our study, there were signi cant racial/ethnic differences in the amount of breast pain reported in a multivariate analysis, where Chinese, Native Hawaiian, and Mixed Asian participants reported signi cantly less pain compared to White participants on a 10-point pain scale. The literature is sparse on breast pain differences according to different racial/ethnic groups. In related literature, studies have reported an increased breast density in Asian women compared to other races 24,25 . It is not known if this plays a role in intrinsic pain sensitivity or how this may contribute to breast pain after a breast cancer diagnosis. There are literature describing ethnic differences in other types of pain however. In one study looking at knee osteoarthritis patients, Asian American participants had signi cantly higher levels of clinical pain intensity and experimental pain sensitivity than Non-Hispanic White participants 26 . Some studies have evaluated chronic musculoskeletal pain systematically and found differences in pain beliefs, cognitions, and behaviors in patients from different racial backgrounds 27 .
The in uence of race/ethnicity on various aspects of pain sensitivity and the crucial impact that culture has on illness beliefs 27 have been studied. Some have hypothesized that Asian cultures view the inability to tolerate pain as a weakness 28 which may contribute to the lower pain value noted in most races other than White in our population. Other studies show a variable range in pain threshold and tolerance among different ethnic groups 17 . There is some evidence showing the differences in coping strategies and illness perceptions between different populations 27 . In one study looking at patients in Singapore 29 , they were less likely to seek social support and to use guarding and resting to cope with their pain. There has been the suggestion that stoicism is a positive trait 30 among Asian cultures, suggesting potential positive reinforcement of certain coping strategies, in addition to whether the reporting of pain would be viewed as favorable or unfavorable and how that would come to play in a clinical environment. There are many other important factors that are not mentioned here, in regards to ethnic group, the ethnic group of the provider and whether that is the same as the patients, and many other important factors, as well. The racial/ethnic diversity of our demographic adds to the existing literature and identi es unique differences in the amount of breast pain experienced and suggests potential explanations of these differences, while we also acknowledge that we are providing only some of the many hypotheses that are relevant and need to be explored.
Multiple studies have evaluated a number of potential risk factors for developing breast pain after breast cancer diagnosis including: age 2,6,8−10 , type of surgery 10 , axillary lymph node dissection 2,4,6,8−10 , use of chemotherapy 4,9 and use of radiation therapy 4,31 . Similar to prior studies, we report that younger patients are more likely to have and/or report breast pain compared to older patients. Breast density is a known risk factor for breast cancer development and survival 32 where younger women are more likely to have dense breasts compared to older women 33 . There is little known about breast density and connection with breast pain. Most of the literature discusses breast density and breast pain in the context of hormonal intervention 34 , where there is no change in the incidence of breast pain or breast density based on the hormonal intervention studied. A potential mechanism for increased breast pain among younger women could be their inherent increased breast density. Another hypothesis related to age is that younger women, in more recent years, have been asked about breast pain and encouraged to report pain.
Breast pain differences were seen according to some treatments in our study, speci cally those who received chemotherapy, radiation and endocrine therapy.
However, we found the use of chemotherapy and radiation therapy was associated with decrease in breast pain, contrary to prior studies that have shown an increase in breast pain with these treatment modalities 2,4,9 . We did see an increase in breast pain for those who received endocrine therapy which has been seen in a recent review evaluating common breast pain treatments 35 . However, we did not see any breast pain differences according to the type of surgical management-lumpectomy vs mastectomy and type of axillary lymph node management, contrary to prior data reported 2,10 . There is some variability with prior studies, where some studies have found breast pain to be independent from clinical factors 36 . Based on a recent review, the variation in disease, treatment mode and symptom progression in various studies make it di cult to draw conclusions about chronic pain from both the prospective and retrospective literature 37 . The literature has focused predominantly on risk factors for the development of chronic pain, however some studies have evaluated potential chronic pain mechanisms which may be an important key to determine speci c treatments. Studies have evaluated genetics variations in cytokine and potassium channel genes to identify persistent breast pain phenotypes 38,39 and their potential mechanisms 40 . These are interesting ndings, however their translational merit has yet to be realized.
We found that sensory qualities of pain were more commonly reported than affective characteristics overall. According to the sfMPQ sensory qualities of pain include: throbbing, shooting, stabbing, sharp, gnawing, cramping, hot, aching, heavy, tender and splitting, and the affective pain quality descriptors include: tiring-exhausting, sickening, fearful and punishing-cruel. These various qualities of pain help to identify characteristics of pain and suggest potential treatments which can address these symptoms uniquely. Depending on the group descriptor that is identi ed, a speci c treatment modality could be implemented. For example, a descriptor that falls within the group category of sensory such as tender or splitting, are characteristics of neuropathic type pain and may be more amenable to neuropathic directed treatments with therapeutics such as gabapentin. Whereas the descriptor sharp which is also sensory but in the incisive pressure group are characteristic of mechanical pain and may be more responsive to direct physical manipulation with massage. Similarly, affective pain quality descriptors may indicate an emotional component to the pain and suggest psychologically focused treatment. In our study, patients with breast pain reported other chronic pain, most commonly lower extremity (LE), followed by back and upper extremity (UE). Unfortunately, chronic pain is common in breast cancer survivors where some studies have seen reported up to 84% reporting chronic pain with and 97% experiencing pain at least 1-3 days/week 41 .
The pain experience is complex. Our current ndings suggest that racial differences are not related to cancer treatments, rather, more likely related to differences in pain perception and coping strategies. Other minority groups compared with Whites have been previously described as reporting and/or experiencing pain differentially, however in Asian Americans, pain remains poorly understood. The sources of pain disparities among racial and ethnic minorities are complex, involving patient (e.g., patient/health care provider communication, illness perceptions, treatment preferences for pain), health care provider (e.g., false beliefs), and health care system (e.g., access to pain medication) factors. There is limited data regarding racial and ethnic differences in patients' treatment-related symptom management experiences. Pain has signi cant socioeconomic, health, and quality-of-life implications. The patient's experience of their illness and its impact on the perception of persistent pain may shed some light on more bene cial outcomes. Further prospective and longitudinal studies using standard de nitions for race, ethnicity or culture and valid questionnaires for each population are warranted to explore the racial, ethnic and cultural discrepancies in pain beliefs, cognitions, and behaviors. A comprehensive pain research agenda is necessary to address pain disparities among racial and ethnic minorities.
In our study, the most common treatment modalities reported to help relieve breast pain were manual therapy, followed by medication and thermal application, although there were a variety of responses. Physical therapy has demonstrated e cacy in treating pain, in some speci c cancers or in treatments sequelae (ie. Lymphedema) 42 . Studies have suggested utilizing rehabilitation/physical therapy to treat multiple neuromuscular and musculoskeletal pain from breast cancer treatment 43 . Of the traditional analgesic medications, non-steroidal anti-in ammatory drugs were most commonly used in our study. A recent study showed that regional blocks are more conclusively effective than pharmacologic modalities for patients after undergoing surgery 44 . For those who may not nd relief with over the counter medications, this study suggests chronic breast pain may be treated with a similar strategy already utilized for other pain postoperatively.
Most complementary and integrative therapies only showed trends in e cacy, as cancer pain is mainly a secondary endpoint in most studies. Due to the multimorphism of cancer pain, certain mind-body therapies such as massage, acupuncture, healing touch, hypnosis, and music therapy can help to address anxiety, stress, depression, or mood disturbances. Other therapies such as yoga, tai chi/qigong, guided imagery, virtual reality, and cognitive-behavioral therapy alone or combined, have shown trends in reducing the severity of cancer pain 42 . In several recent randomized controlled trials, hypnosis positively in uenced pain, distress, fatigue, and nausea 45 . Mindfulness-based intervention has been found e cacious in reducing persistent pain in women treated for breast cancer 46 . In a recent review of various pain therapies, studies were varying where interventions most often include education and relaxation techniques, followed by interventions with behavioral or cognitive components 47 . There remains a need for research of current psychotherapeutic interventions and their e cacy and the role of mediator variables (e. g., coping) on pain perception in cancer patients.
Racial and ethnic minorities tend to be undertreated for pain when compared with non-Hispanic Whites 48,49 . These racial and ethnic disparities in pain perception, assessment, and treatment have been found in several clinical settings (i.e., postoperative, emergency room) and types of pain (i.e., acute, cancer, chronic nonmalignant, and experimental) 48 . Given the anonymous nature of our assessments, we anticipated a more accurate representation of breast pain in our population. Prior to initiating this study, we asked clinicians if breast pain was a problem in the community. A majority of providers (surgeons, oncologists, radiation oncologists) said that their patients have never reported breast pain, with a minority of clinicians saying it has come up at times. When asked if they assessed for breast pain, a majority of physicians reported they did not. In a study looking at clinicians' beliefs, including medical students and residents 48 , found that false beliefs about racial/ethnic biologic differences could contribute to disparities in pain assessment and treatment 50 . By exploring the racial, ethnic and cultural discrepancies in pain beliefs, cognitions, and behaviors, a shared understanding between the patient and professional could be developed, likely leading to both the elucidation and better treatment for breast pain.

Limitations
This was a questionnaire based study and a subjective self-report measurement of pain was used to identify breast pain. One limitation was participant recall bias. The rst survey question regarding pain was limited to those who have experienced pain in the last 6 months. However, this remains a major limitation of this study. This assessment was limited to English speaking patients and excludes a small but signi cant percent of our patient population. Validating the questionnaire in other languages will be important in capturing the totality of experiences within our state. Third, we were unable to ascertain the etiology of the pain problems faced by the study participants. It is possible that relationships between pain beliefs/coping and adjustment to pain may be moderated, at least partly, by pain etiology. The cross-sectional nature of the study precludes us from drawing conclusions regarding the potential impact on pain-related outcomes. Despite the study's limitations however, the ndings provide preliminary data that can lead to suggestions of potential cultural-speci c interventions to address breast pain symptoms.

Conclusions
Breast pain is a signi cant symptom in our breast cancer community. This questionnaire has informed our understanding of the type of pain our diverse breast cancer patients are experiencing and in turn we are developing culturally appropriate pain management strategies to treat this challenging symptom for breast cancer survivors. Due to the complexities of the pain experience, identifying key characteristics of pain is important in order to provide appropriate therapy. Utilizing complementary and integrative therapies can help to address cancer pain. However, implementing these therapies requires a thorough understanding of the nature of the pain and an interdisciplinary team approach to offer personalized, cancer pain management, as many aspects of the cancer experience, such as provider patient communication, has relevance in this area of work. Elucidating ethnic group differences has translational merit for culturally competent clinical care and for addressing and reducing pain treatment disparities among ethnically/racially diverse groups 17 . We are currently preparing an interventional study to address breast pain in our population, utilizing what we have learned from this study.

Declarations
Ethics approval and consent to participate The Institutional Review Boards at the University of Hawaii and the individual Hospital Research Institutes approved this study. All aspects of the investigation complied with current rules for research involving Human Subjects. The need for verbal or written consent was waived and deemed unnecessary according to the University of Hawaii IRB. A written consent form was provided to each participant which explain the purpose of the study and anonymous nature. By completing the online survey or returning the completed paper survey, participants imply their consent to participate in the study.

Consent for publication
Not applicable Availability of data and material The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no completing interests.   Breast Pain Characteristics. Overall, the most common descriptors of breast pain were from sensory qualities of pain compared to affective pain quality descriptors. For mild pain; aching, for moderate pain; sharp and there was not a most common descriptor for severe pain.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. bpainquestionaire.docx STROBEchecklistJFBreastpain.doc