Social and clinical drivers of stress responses in African American breast cancer survivors

Abstract

Racial differences in breast cancer morbidity and mortality have been examined between Black/African American women and White women as part of efforts to characterize multilevel drivers of disease risk and outcomes. Current models of cancer disparities recognize the significance of physiological stress responses, yet data on stress hormones in Black/African American women with breast cancer and their social risk factors are limited. We examined cortisol levels in Black/African American breast cancer patients and tested their association with social and clinical factors to understand the relationship between stress responses and women’s lived experiences. Seventy-two patients who completed primary surgical treatment were included in this cross-sectional study. Data on sociodemographic characteristics and chronic diseases were obtained by self-report. Breast cancer stage and diagnosis date were abstracted from electronic health records. Cortisol levels were determined from saliva samples. Compared to those without hypertension, patients with hypertension were 6.84 (95% CI 1.33, 35.0) times as likely to have high cortisol (p = 0.02). The odds of having high cortisol increased by 1.42 (95% CI 1.03, 1.95, p = 0.03) times for every point increase in negative life events. Hypertension and negative life events are associated with high cortisol levels in Black/African American patients. These findings illustrate the importance of understanding the lived experiences of these patients to enhance cancer health equity.

Introduction

Breast cancer is one of the leading cancer sites among women in the United States that disproportionately affects Black/African American women in terms of morbidity and mortality^1–3. In addition to racial differences in morbidity and mortality, national data have now shown a convergence in breast cancer incidence between Black/African American women and white women^4–6. With mortality rates that continue to be higher among Black/African American women^6–8, there is the potential for even greater racial disparities from breast cancer. For this reason, efforts are now focused on improving the precision of early detection, prevention, and treatment among Black/African American women by identifying multilevel drivers that contribute to increased risk and poorer outcomes. An emerging hypothesis about breast cancer disparities is that one’s lived experience and physiological responses to social stressors influence biological processes that are involved in the disease’s initiation and progression^9,10. For instance, findings from animal research have provided important insights about the biological impact of exposure to certain types of social stressors on the development of mammary tumors¹¹ and studies are beginning to examine the link between exposure to social stressors (e.g., racial discrimination, living in disadvantaged neighborhoods) and racial disparities in breast cancer risk and outcomes^12–15. In one report, Hermes and colleagues¹¹ found that exposure to social isolation was associated with an increased corticosterone response among rats, and animals who had a dysregulated cortisol response were more likely to develop mammary tumors that are histologically similar to those that develop among Black/African American women.

Cortisol is the primary hormone associated with stress in humans and represents the end product of the hypothalamic–pituitary–adrenal (HPA) axis activation, which occurs in response to environmental, psychological, or physiological stressors¹⁶. Cortisol receptors are present in almost all cell types¹⁷ and this hormone plays a critical role in metabolism, inflammatory responses, and overall immune functioning^16,18,19. Cortisol levels can be used to characterize HPA-axis functioning and understand physiological responses to stressors²⁰. Evidence is emerging about the links between abnormal cortisol production, activation of tumorigenic pathways that are involved in the initiation and progression of disease, and HPA-axis dysregulation^21,22. For instance, previous studies have shown that African Americans, and those who have greater exposure to socioeconomic stressors (e.g., low income), are likely to have a dysregulated cortisol response^23–26. Other studies conducted in Europe reported associations between increase cortisol and relapse²⁷ as well as increase cortisol and lower quality of life²⁸ in white women with a history of breast cancer. However, little is known about cortisol levels specifically among African American women who have a personal history of breast cancer even though these women are likely to have greater exposure to adverse social conditions¹⁸ that are associated with poorer survival¹². There is a critical need to identify Black/African American women who have a personal history of breast cancer who may exhibit dysregulated HPA axis functioning, characterized by either high or low cortisol levels, at a critical timepoint in the progression from diagnosis to treatment and its possible effect on the cancer outcomes. To address this gap, we measured cortisol levels in a sample of African American breast cancer patients and also examined the association between cortisol levels and socioeconomic and social stressors that have been explored in mechanistic studies of breast cancer disparities⁹. We predicted that patients who have greater socioeconomic stressors (e.g., low income), social isolation, and negative life events would have higher cortisol levels.

Materials and methods

Study design and sample

This is a cross-sectional observational study of baseline data conducted as part of a trial with Black/African American breast cancer patients that was conducted at an NCI-designated cancer center in South Carolina. Participants in this study had been diagnosed with early-stage or locally advanced (stage I through IIIa disease) between 2018 and 2022. Patients were eligible to participate in the study if they completed primary surgical treatment and were ages 21 through 75 years when they were diagnosed. The average amount of time from diagnosis to study enrollment was 2.8 (SD = 1.5) years. The study was approved by the Institutional Review Boards at the Medical University of South Carolina and the University of Southern California. All methods were performed in accordance with relevant guidelines and regulations.

Procedures

The procedures for this study have been described in detail previously²⁹ and are summarized here. Given that the present study focuses on testing a secondary hypothesis using baseline data of study participants collected prior to the trial, a power calculation is not applicable and not included in this report³⁰. Patients were recruited to participate in the study using two strategies. In the first strategy, Black/African American breast cancer patients were identified from the tumor registry at an NCI-designated cancer center and were mailed an introductory letter that explained the purpose of the study, the procedures involved in participation, and a flyer that provided information about the study investigator. These materials described the study as a project that was designed to develop a better understanding of how Black/African American women cope with their breast cancer diagnosis and treatment. Patients were directed to contact the study team within two weeks using a toll-free number or email if they were not interested in being contacted about the study. Patients who did not opt out of participation were contacted by a research assistant by telephone to review the introductory packet, obtain verbal informed consent, and complete a brief screening interview to determine eligibility and complete the baseline telephone interview described below. The second recruitment method was an in-person clinic strategy in which potential participants were identified by screening the breast cancer clinic schedules and reviewing electronic medical records to identify patients who met the diagnostic and treatment parameters described above. Eligible patients were informed about the purpose and course of the study procedures during a clinic visit using the informed consent form as a guide. Patients who were interested in participating in the study provided written informed consent and scheduled a time to complete the baseline telephone interview.

A 30-min baseline telephone interview was completed with patients following the study invitation and provision of verbal informed consent to obtain data on socioeconomic, social, and psychological stressors by self-report. At the end of the baseline telephone interview, patients were invited to complete two laboratory visits to measure clinical biomarkers (e.g., blood pressure, heart rate) and cortisol levels at the pre-challenge visit and to examine reactivity to a laboratory-based stressor during a stress exposure visit. During the pre-challenge visit, saliva samples were collected at 3:00 PM and no later than 3:15 PM to control for diurnal variations in cortisol³¹. The current study focuses on measures of levels of salivary cortisol obtained during the pre-challenge visit to provide insight into HPA axis functioning within the context of a typical day for these patients.

Measures

Socioeconomic stressors

Self-reported data on marital status, education level, employment status, and income were obtained during the baseline telephone interview using items from our previous research²⁹. Responses to these items were re-coded into dichotomous variables to reflect greater exposure to socioeconomic stressors (e.g., low income, unemployed) based on the distribution of responses.

Clinical factors

Breast cancer stage, month/year of diagnosis, and disease stage were abstracted from electronic health records. Hypertension status was obtained by self-report during the baseline telephone interview by asking “Please tell me if a doctor or health care provider has ever told you that you have high blood pressure”. Patients were categorized as having a personal history of hypertension or not having this disease (yes versus no).

Salivary cortisol

A Salimetrics purple top salivette was used to collect saliva specimens. Saliva samples were collected at 3:00 PM, and no later than 3:15 PM. Samples were collected by swabbing the cheek of the patient for at least 2 min. The swab was then returned to the collection tube, capped, and placed in a cup of ice until the end of the visit and taken to an institutional laboratory facility for processing, analysis, and storage. Collected samples were frozen at − 20 °C or below for a minimum of 2 h, thawed, and centrifuged prior to laboratory testing. Salivary cortisol was assessed using the Tecan Cortisol Saliva Luminescence Immunoassay. One individual sample was excluded for being an outlier (Salivary cortisol = 9.8 mg/dl, which is over 3 standard deviation beyond the mean value of 0.275 mg/dl).

Social stressors

The Life Events Questionnaire (LEQ)^32,33 was used to determine the number and impact of life events during the past year^33,34. Specifically, participants were asked to indicate events related to their health (e.g., major illness or injury), work (e.g., difficulty finding a job), financial (e.g., major change in finances), or crime/legal matters (e.g., being robbed or victim of identity theft) was experienced during the past year, if these were good or bad events, and the effect of the event on their life (no effect, some effect, moderate effect, great effect). A negative events score was calculated by summing the impact ratings for items that participants reported as being bad events³³. Perceived stress was assessed using the abbreviated 4-item Cohen Perceived Stress Scale (PSS) (e.g., unable to control the important things in your life)³⁵, and perceived social isolation or loneliness (e.g., how often do you feel isolated from others?) was measured using the UCLA 3-item Loneliness scale questionnaire³⁶. The Cronbach’s alpha for PSS and social isolation were 0.70 and 0.77 respectively. Higher scores on these measures indicated greater levels of perceived isolation, perceptions of stress, and more frequent or severe negative life experiences.

Data analysis

First, descriptive statistics were generated to characterize the study sample in terms of sociodemographic characteristics, clinical variables, and cortisol levels. Next, cortisol was re-coded into a dichotomous variable using the median value to measure high versus low cortisol levels. Next, bivariate analyses were conducted to examine association between cortisol levels and sociodemographic and clinical characteristics. For categorial variables, the Pearson Chi-Square test was used to compare the proportions of participants with high vs. low cortisol levels. For continuous variables, the independent two-sample t-test was used to compare means across groups with high vs. low cortisol values. This approach was used to align with the methods used to calculate allostatic load³⁷ and a strategy that could be implemented into clinical practice. Dichotomization also helps interpret the impact of elevated cortisol on health outcomes and addresses the skewed distribution of cortisol levels. Last, multivariable binary logistic regression analysis including all regressors simultaneously was conducted to identify factors having significant independent associations with cortisol levels. BMI, known to be associated with cortisol levels^38,39, and variables that had a p < 0.20 association with cortisol levels were included in the multivariable logistic regression model. This p-value was used as the criteria for selecting variables for inclusion in the multivariable logistic regression model because limited empirical data are available on cortisol levels specifically among Black/African American women.

Results

Of the 110 patients who were enrolled in the study, 72 (65%) completed the pre-challenge visit and had complete cortisol data. Table 1 shows the characteristics of the study sample and provides descriptive information on socioeconomic and social stressors. The majority (61%) of patients were not married, had at least some college education (78%), and were 50 years or older (82%). Forty-seven percent of patients were employed and 44% had an income greater than or equal to $35,000. With respect to clinical characteristics, 57% of patients had been diagnosed with breast cancer more than two years ago, 50% had stage 1b or lower disease, and 71% of women had hypertension.

Table 1.

Sample Characteristics and Bivariate Analysis of Cortisol (n = 72).

Variable	Level	n (%)	N (%) High Cortisol	p-value
Marital status	Married	28 (39)	9 (32)	0.02*
	Not Married	44 (61)	27 (62)
Education level	≥ Some College	56 (78)	26 (46)	0.26
	≤ High School	16 (22)	10 (63)
Employment status	Employed	34 (47)	15 (44)	0.63
	Retired	23 (32)	13 (57)
	Not Employed	15 (21)	8 (53)
Income level	≥ $35,000	30 (44)	10 (33)
	< $35,000	38 (56)	25 (68)
Time Since diagnosis	≤ Two Years	31 (43)	16 (52)	0.81
	> Two Years	41 (57)	20 (49
Stage	IIa or greater	36 (50)	17 (47)	0.64
	Ib or lower	(36 (50)	19 (53)
High blood pressure	Yes	51 (71)	32 (63)	< 0.01*
	No	21 (29)	4 (19)
BMI	≥ 75th percentile	16 (22)	10 (63)	0.26
	< 75th percentile	56 (78)	26 (45)
Age	≥ 50 years	59 (82)	33 (56)	0.03*
	< 50 years	13 (18)	3 (23)

Variable	Total Sample Mean (SD)	High cortisol mean (SD)	Low cortisol mean (SD)	p-value
Age	58.3 (8.7)	59.5 (8.4)	57.1 (9.1)	0.24
Perceived social isolation	4.6 (1.8)	5.1 (2.0)	4.3 (1.6)	0.05
Negative events	3.9 (2.5)	4.4 (2.8)	3.3 (2.2)	0.06

BMI body mass index.

p-value for categorical variables obtained using Pearson Chi-Square test and for continuous variables using independent two sample t.test.

*Statistically significant p-value =  < 0.05.

sample may not equal n because of missing data.

The mean (SD) and median (IQR) levels for cortisol were 0.275 (1.14) mg/dl and 0.11 (0.07) mg/dl respectively. As shown in Table 1, patients who were not married (p = 0.02), those who had annual income levels less than $35,000 (p = 0.01), and patients who were ages 50 and older were most likely to have high cortisol levels. Similarly, patients who had a personal history of high blood pressure (p < 0.01) were also most likely to have a high cortisol value compared to those who did not have hypertension (63% versus 19%) In addition, levels of social isolation were significantly greater among patients who had high cortisol levels (Mean = 5.1, SD = 2.0) compared to those who had lower cortisol (Mean = 4.3, SD = 1.6) (t = 1.97). In addition, negative life events were higher among patients who had high (Mean = 4.4 (SD = 2.8) versus low (Mean = 3.3, SD = 2.2) (t = 1.88) cortisol levels.

Table 2 shows the results of the multivariable logistic regression and odd ratios for predictors of high cortisol. Hypertension status and negative life events had statistically significant independent associations with high cortisol levels. Specifically, patients who had hypertension were 6.84 (95% CI = 1.33, 35.0, p = 0.02) times as likely to have high cortisol than those without hypertension. The odds of having high cortisol also increased 1.42 (95% CI = 1.03, 1.95, p = 0.03) times for every point increase in negative life event score. Although not statistically significant, the odds of having high cortisol in those with BMI greater than 75th percentile were 5.05 (95% CI = 0.83, 30.4, p = 0.08) times the odds of those with normal BMI.

Table 2.

Multivariate Logistic Regression Model for High Cortisol.

Variable	Level	Odds Ratio	95% confidence interval	P-value
Income level	≥ $35,000 (Ref)  < $35,000	2.15	0.53, 8.81	0.29
Marital status	Married (Ref) Not Married	2.77	0.71, 10.9	0.14
High blood pressure	No (Ref) Yes	6.84	1.33, 35.0	0.02*
Social isolation	N/A	1.17	0.79, 1.73	0.43
Negative life events	N/A	1.42	1.03, 1.95	0.03*
BMI	< 75th percentile (Ref)  ≥ 75th percentile	5.05	0.84, 30.4	0.08
Age	< 50 years (Ref)  ≥ 50 years	2.14	0.26, 17.9	0.49

Odd ratios obtained using multivariable logistic regression. Variables mutually adjusted for each other.

Ref = Reference group.

*Statistically significant p-value =  < 0.05.

N/A indicates continuous variable.

Discussion

There is substantial empirical data about racial disparities in breast cancer outcomes among Black/African American women^2,3,15, and efforts are now being made to understand the social context of breast cancer diagnosis, treatment, and recovery among women who have a personal history of disease⁴⁰. To our knowledge, this is the first study to examine the physiological stress response, as measured by cortisol, among Black/African American women breast cancer survivors and to identify clinical and social factors having significant independent associations with cortisol levels. We found that hypertension status and negative life events had significant independent associations with high cortisol levels; patients who reported a personal history of hypertension were 6.84 (95% CI = 1.33, 35.0, p = 0.02) times as likely to have high cortisol levels. The human stress response is regulated by two major physiological systems: the autonomic nervous system and the HPA axis⁴¹. Stress-related HPA activation results in cortisol secretion⁴². Prolonged exposure to cortisol can adversely impact metabolic functions and is associated with the development of conditions such as obesity⁴², diabetes⁴³, and cardiovascular disease⁴⁴. Importantly, cortisol plays a role in regulating blood pressure⁴⁵; thus, the positive association between cortisol and hypertension status is consistent with this regulatory function.

An emerging hypothesis about racial disparities in breast cancer among African American women is that social conditions and physiological responses to social stressors influence biological processes that are important to the initiation and progression of the disease¹¹. In addition to experiencing the physical and financial stressors associated with breast cancer treatment, Black/African American women may also be exposed to chronic economic stressors (e.g., low income, financial strain) and live in medically underserved geographic areas that are associated with being diagnosed with higher stage breast cancer⁴⁶. Notably, 56% of patients in this study had an annual household income of less than $35,000, and 61% were not married. Among breast cancer patients in South Carolina (the geographic location for this study), being unmarried was associated with a greater likelihood of distant stage at diagnosis⁴⁶.

In the present study, perceived social isolation was associated with high cortisol levels in the bivariate analysis, but only negative life events had a significant independent association with cortisol in the multivariable logistic regression analysis. Similarly, income level had a significant association with cortisol in the bivariate analysis, but this association was attenuated when considered together with negative life events in the multivariate logistic regression model. Patients who had greater negative life events had an increased likelihood of having high cortisol levels. To our knowledge, our study is the first to demonstrate a positive association between cortisol levels and self-reported negative life events in Black/African American breast cancer patients. Negative life events (e.g., employment issues, interactions with the legal system) reflect the cumulative adverse effects of social stressors. Taken together, these findings suggest that the overall negative life experiences, rather than individual socioeconomic indicators of adversity are associated with stress response markers. Thus, as efforts are being made to understand the association between breast cancer risk factors and place-based measures of structural racism (e.g., redlining)²⁷ and residence in geographic areas that have social deprivation^12,47,48, it is important to also understand the actual lived experiences among cancer patients and survivors. The findings from the present study underscore the importance of obtaining multilevel data on lived experiences using self-report data and other types of measures.

This study presents several strengths including the objective measurement of cortisol levels as an outcome, and its focus on Black/African American breast cancer survivors, an underrepresented group in research. In considering the results of this study, several limitations should be considered. These include the cross-sectional analysis of cortisol levels in a relatively small sample of Black/African American breast cancer patients who were included in this observational study. Specifically, salivary cortisol was sampled at a one-time point at the pre-challenge visit that was completed in the afternoon. This method prevented us from measuring cortisol at other critical time points (e.g., morning cortisol awakening response) or evaluating diurnal cortisol slope (DCS) changes⁴⁹. Future studies should explore the relationships between diurnal cortisol patterns and cortisol at other time points in African American breast cancer patients to further investigate HPA functioning in this population. Nonetheless, higher levels of afternoon salivary cortisol may suggest greater exposure to total daily salivary cortisol, and a blunted DCS⁵⁰, which has been associated with lower breast cancer survival^51,52. The limitations of our study design and sample size are offset by the specific inclusion of Black/African American women who have a personal history of breast cancer. The challenges associated with recruiting diverse cancer patients to participate in research, especially prospective studies, are well-documented⁵³. In larger breast cancer cohort studies that were developed over multiple years^18,54,55, Black/African American women made up less than 10% of the sample. The limitations of the sample size in our study may have lowered our statistical power and contributed to the wide confidence intervals for some of the associations in the regression⁵³ Additional research is needed to examine cortisol levels with larger samples of Black/African American breast cancer patients using prospective study designs. Nevertheless, our study provides novel empirical data about stress responses among a priority population for cancer health disparities and illustrates the importance of understanding the lived experiences of female Black/African American breast cancer survivors as part of efforts to enhance cancer health equity. Our findings underscore the importance of screening patients for social determinants of health in both in-patient and ambulatory settings. Further, it may be important to measure cortisol as part of cancer care delivery. Future research is needed to examine the downstream effects of high cortisol levels on Black/African American breast cancer patients and other groups that are at increased risk for poor outcomes following diagnosis and treatment.

In conclusion, our findings highlight the importance of adopting a whole-person health approach in minority health, health disparities, and efforts that are made to enhance equity in health care. This involves not only improving the precision of health care using therapeutic strategies, but also considering the patient's lived experiences and social stressors. As part of this, efforts are needed to enhance the quality and impact of screening and referral for social determinants of health in clinical settings.

Author contributions

CHH contributed to the conception, design of the work, analysis, and writing of the manuscript. FS contributed to the design, statistical analysis, and writing of the main manuscript text. MJ contributed to project data/management, data cleaning, and writing the manuscript TB, OAB, and BS contributed to the manuscript through substantial revisions and editing. JC contributed to statistical methods and drafted key sections of the manuscript.

Funding

This study was supported by National Institutes of Health (NIH) Science of Behavior Change Common Fund Program through awards administered by The National Cancer Institute (R21CA235852 and U54CA233465. This work was also supported by U54MD010706 from the National Institute on Minority Health and Health Disparities and the U54CA233465 Award from NCI.

Data availability

The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Competing interests

The authors declare no competing interests.