Factors influencing Patient Preferences for BRCA Testing and Adjuvant Therapy in HER2-Negative Early Breast Cancer in the United States: Best–Worst Scaling and Discrete Choice Experiment

Abstract

Introduction

Poly(ADP-ribose) polymerase inhibitors (PARPi) have survival benefits for patients with high-risk (High-risk disease is defined per the phase III OlympiA trial as follows: for triple-negative breast cancer, residual disease after neoadjuvant chemotherapy or node-positive or ≥ 2 cm tumors after adjuvant chemotherapy; for hormone receptor-positive disease, four or more positive nodes after adjuvant chemotherapy or a CPS + EG score ≥ 3 after incomplete response to neoadjuvant chemotherapy. The CPS + EG score accounts for clinical/pathologic stage, ER status, and grade (Giaquinto et al. in CA Cancer J Clin 72:524–541, 2022)), human epidermal growth factor receptor 2 (HER2)-negative early breast cancer (eBC) with germline BReast CAncer gene mutations (gBRCAm). However, many patients are unaware of their gBRCA status; this can impact eligibility for targeted treatment. We sought to evaluate patient preferences for BRCA testing and treatment decision-making as they relate to HER2-negative eBC.

Methods

We conducted an online survey, including a best–worst scaling exercise (BWS) and discrete-choice experiment (DCE), among patients with self-reported HER2-negative eBC residing in the USA who were either untested, unsure if they were tested, or tested positive for the gBRCAm. The BWS generated a rank ordering of 16 barriers and facilitators to BRCA testing. The DCE evaluated patient preferences for adjuvant therapies versus no treatment based on seven treatment attributes: invasive disease-free survival, targeted treatment, nausea risk, risk of serious side effects, regimen, treatment duration, and cost. BWS and DCE exercises were analyzed using hierarchical Bayesian models.

Results

Among the 359 women included in our sample, the top facilitators for BRCA testing were determining eligibility for targeted therapy that may prevent or delay metastasis, a physician’s recommendation, and absence of out-of-pocket costs (OOPC). In contrast, the top barriers were an OOPC of $250, potential anxiety from test results, and the possibility of a 3- to 4-week delay in treatment. The DCE showed that most participants preferred adjuvant treatment (77.6%) over no treatment, and reducing treatment OOPC from $900 to $0, reducing the risk of serious side effects from 77 to 24%, and having a BRCA-targeted treatment influenced treatment choice most.

Conclusions

Individuals reported that a key benefit of BRCA testing was the insight it provided into their treatment options, allowing for more personalized care. However, OOPC was a barrier to testing. Their choice to receive adjuvant therapy was most influenced by OOPC, followed by the tolerability of the treatment and the ability to receive a targeted therapy.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40271-025-00783-1.

Plain Language Summary

Poly(ADP-ribose) polymerase inhibitors (PARPi) are a type of cancer medicine that kills cancer cells by stopping them from fixing their DNA. PARPi are helpful for people with early breast cancer (eBC) who are more likely to have their cancer come back, who do not have the human epidermal growth factor receptor 2 (HER2) protein (called HER2-negative), and have a change in certain genes called BRCA1 or BRCA2. These gene changes make it more likely someone will get breast cancer. PARPi can help people live longer and keep the cancer from coming back. But many people do not know whether or not they have a BRCA gene change. This study aimed to find out what is most important to people with HER2-negative eBC when it comes to BRCA testing and choosing treatments. A total of 359 women with HER2-negative eBC took an online survey. Some had never been tested for BRCA, some were not sure whether they had, and some tested positive. The survey looked at why people would or would not get BRCA testing. It also looked at whether they would want more treatment after surgery and why people would choose treatment. A main reason for getting BRCA testing was to see whether they could get a treatment that might stop or slow down the cancer from spreading. Cost was a main reason for not getting testing. Many people wanted treatment after surgery rather than no treatment. Cost was most important, followed by risk of serious side effects, and whether the treatment was targeted based on a BRCA change.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40271-025-00783-1.

Key Points for Decision-Makers

Most individuals are open to adjuvant therapy and are motivated to consider BRCA testing to inform eligibility for targeted adjuvant treatment. This underscores the importance of determining a patient’s BRCA status to select the most effective therapy.

Out-of-pocket costs are a key factor influencing a patient's decision about undergoing BRCA testing. Refusing testing due to cost prevents access to BRCA-targeted therapy, which may lead to poorer outcomes and higher long-term costs for the payer.

Minor differences in patient preferences were observed for adjuvant therapy and BRCA testing, highlighting the importance of considering individual patient preferences when selecting an adjuvant treatment.

Introduction

Breast cancer is the second most common cancer in the USA, with a 5-year prevalence of 1.8 million patients and approximately 340,000 new cases diagnosed each year [1, 2]. Despite advances in early detection and treatment, breast cancer is the second leading cause of cancer-related death among women [1]. From a treatment perspective, breast cancer is classified into subtypes based on the presence of hormone receptors (HR) and human epidermal growth factor receptor 2 (HER2). The four main subtypes are HR+/HER2-negative, HR-negative/HER2-negative (also known as triple-negative breast cancer), HR+/HER2+, and HR-negative/HER2+, with 81% of all breast cancer cases classified as HER2-negative [3].

In the USA, over 90% of new breast cancer diagnoses occur at an early stage, when the disease is confined to the breast or regional lymph nodes [4]. Treatment for early breast cancer (eBC) typically includes surgical resection with curative intent, often followed by adjuvant therapies, including chemotherapy, endocrine therapy, or targeted therapies [5]. In select patients, neoadjuvant therapies, such as chemotherapy, targeted therapy, immunotherapy, or endocrine therapy, may be administered to reduce tumor burden and improve surgical outcomes [5]. For HER2-negative eBC, the treatment landscape has rapidly evolved, with new approvals for immunotherapy, cyclin-dependent kinase (CDK) 4/6 inhibitors, and a poly(ADP-ribose) polymerase inhibitor (PARPi) expanding options for adjuvant and neoadjuvant therapy [6–9]. Among these advances is olaparib, a PARPi approved for patients with germline BReast CAncer mutations (gBRCAm) in the BRCA1 or BRCA2 genes, introducing further biomarker segmentation in this subtype [9]. These advances have made treatment choices more complex, creating a need to understand patient preferences in this changing therapeutic environment.

BRCA1 and BRCA2 genes are tumor suppressor genes involved in the homologous recombination repair of DNA damage. Mutations in these genes compromise the ability to repair DNA, resulting in increased genomic instability and an elevated risk of developing cancer [10]. In the USA and Europe, approximately 5–10% of patients with breast cancer harbor gBRCAm [2]. Historically, undergoing genetic testing to assess BRCAm status was first used to diagnose individuals with hereditary breast and ovarian cancer syndrome, which identify patients who should be prioritized for closer cancer screening [10]. However, it is now used as a critical tool in guiding treatment decisions in eBC [5]. This shift was driven by results from the international, phase III OlympiA trial, which demonstrated that adjuvant olaparib significantly improved invasive disease-free survival, distant disease-free survival, and overall survival compared with placebo in patients with high-risk, gBRCAm, HER2-negative eBC who previously received neoadjuvant or adjuvant chemotherapy [11, 12]. Based on these results, the US Food and Drug Administration approved adjuvant olaparib in March 2022 for use in patients with high-risk, gBRCAm, HER2-negative eBC [9]. Current US guidelines now recommend BRCA testing for all patients with high-risk, HER2-negative eBC to determine eligibility for olaparib, regardless of age or family history [13–15].

Despite guideline recommendations, BRCA testing remains underutilized, limiting access to adjuvant olaparib [16]. Survey data from oncology healthcare professionals and patients in the USA indicate that approximately 18–30% of patients with high-risk, HER2-negative eBC do not undergo BRCA testing, a finding that aligns with a real-world evidence study using US electronic medical record data from May 1, 2020, through May 31, 2022 [17, 18]. Several patient-related barriers to BRCA testing have been identified, including out-of-pocket costs for testing and genetic counseling, perceived lack of physician familiarity with BRCA testing, limited understanding about the role of BRCA testing in guiding treatment decisions, patient concerns and fears, and patient concerns about insurance discrimination [17, 19, 20]. Furthermore, a 2018 survey conducted by the Association of Community Cancer Centers found that 59% of community oncologist respondents reported that patient-related factors, such as fear, refusal, and concerns for future insurability following genetic testing, were barriers to routine BRCA testing [21]. However, most research on BRCA testing uptake and barriers predates the approval of adjuvant olaparib and updated BRCA testing guidelines, limiting its relevance to the current landscape. Understanding these barriers in the context of recent clinical advances is crucial for providing patients with high-risk, HER2-negative eBC access to more personalized treatment options that may improve outcomes.

Given that BRCA mutation status directly influences treatment eligibility for adjuvant olaparib, this interdependence adds complexity to treatment decision-making, in which patients must consider the trade-offs between treatment efficacy, potential side effects, treatment duration, and financial burden. Understanding how patients navigate these decisions is important for informing patient-provider discussions and ensuring treatment choices align with individual patient preferences. Previous studies on patient preferences in HER2-negative eBC within the current treatment landscape have focused on preferences for CDK4/6 inhibitors [22, 23]. To our knowledge, limited research has examined how patients perceive adjuvant olaparib within the context of newly available treatments for high-risk, HER2-negative eBC [24]. The objective of this study was to identify factors that influence patient preferences for BRCA testing and adjuvant treatment decision-making within the current treatment landscape for high-risk, HER2-negative eBC.

Materials and Methods

A cross-sectional, web-based survey was administered to a convenience sample of individuals from the USA with HER2-negative eBC to assess the preferences for gBRCA testing and adjuvant treatment. The survey included a best–worst scaling (BWS) exercise to evaluate how individuals prioritized barriers and facilitators to BRCA testing and a separate discrete-choice experiment (DCE) to evaluate preferences for attributes associated with eBC treatment. BWS results in a rank-ordering of the relative importance of a set of concepts [25], in this case, factors influencing individuals’ decisions to receive BRCA testing. This methodology was appropriate for our study objectives as it permitted the evaluation of the magnitude of the relative importance of key drivers and barriers to BRCA testing. Based on these results, we sought to provide insights for future interventions aimed at increasing the uptake of BRCA testing when indicated. DCEs are useful when assessing trade-offs that individuals are willing to make amongst a set of attributes associated with different treatment options [26]. The attractiveness of a particular treatment to individuals is dependent on their relative preference for treatment attributes evaluated, which is determined based on the frequency with which respondents select product profiles with their preferred attributes. Thus, for this study, DCE was an appropriate methodology for quantifying patient preferences for adjuvant treatments in eBC. We followed the DIRECT reporting checklist for DCEs [27].

Participants

Individuals were included in the study if they self-reported they were ≥18 years of age; residing in the USA at the time of the survey; diagnosed with HER2-negative eBC (stages I–III) in or before 2023; and were either untested, unsure whether they were tested, or positive for the gBRCAm. Because this study was designed to focus on patients who were potentially eligible to receive PARPi to treat their eBC, for patients who had not had BRCA testing, we required that they had already undergone surgery, were scheduled for surgery, or planned to have surgery to treat their breast cancer. Moreover, individuals with confirmed gBRCAm were required to have undergone surgery and received either neoadjuvant or adjuvant chemotherapy (or were planning to receive adjuvant chemotherapy after surgery). Individuals were excluded from this study if they reported having metastatic disease, received anti-HER2 targeted therapy (e.g., trastuzumab, pertuzumab), tested negative for the BRCAm or had “uncertain” test results, or tested positive for the BRCAm and were treated with partial pembrolizumab regimen (defined as having received or planning to receive pembrolizumab either before or after surgery).

Participants were recruited through M3 Global Research, which develops and maintains opt-in patient panels to support healthcare and market research [28]. The sampling frame for this study included both a general population panel and a cancer-specific panel comprising adults who agreed to participate in online surveys. Although using a cancer-specific panel was more efficient for recruiting patients with self-reported cancer diagnoses, adding the general population panel broadened the sampling frame to adults who opted-in to being contacted for online survey research, but who had not yet self-reported that they were diagnosed with cancer. As such, the decision to use both cancer-specific and general population panels was made to maximize both efficiency and diversity within our sample, given the limited size of the gBRCAm patient population. Moreover, with the goal of constructing a diverse analytic sample, soft quotas were monitored during recruitment to ensure a minimum number of individuals in the following categories: gBRCA untested, gBRCAm, Black/African American, Hispanic, previously received neoadjuvant chemotherapy, previously received adjuvant chemotherapy, HR-positive breast cancer, or triple-negative breast cancer.

This study received exemption status from Sterling Institutional Review Board and was conducted according to good research practices for conjoint analyses as recommended by the International Society for Pharmacoeconomics and Outcomes Research [29]. All participants provided electronic informed consent and received fair market value for their participation ($78 for completion of the survey), as determined by the recruitment partner in accordance with limits set by the study sponsor.

Attributes and Levels

Attributes and levels for the BWS and DCE were drafted based on empirical evidence from a targeted literature search and input from clinical experts who were members of the study team. The targeted literature search was conducted to identify prior research examining the barriers to gBRCA testing, including financial burden, as well as factors that may influence treatment selection, such as efficacy endpoints, side effects, and treatment frequency and administration. Study team members with clinical expertise in oncology, including the treatment landscape and biomarker testing processes, provided guidance to support the selection of the most clinically relevant attributes. The final list of attributes and levels for both preference exercises was refined via one-on-one, semi-structured interviews, which occurred between January and February 2023 with 12 patients, 12 oncologists, and eight genetic counselors. Patients included in the qualitative interviews were required to meet the broader study eligibility criteria (described above). Oncologists were eligible for the interviews if they were board certified/eligible in their specialty, spent ≥50% of their professional time engaging in direct patient care, and treated ≥20 patients with eBC in the past 3 months; genetic counselors were eligible if they were certified in their field, practiced for ≥3 years, and advised five or more patients with eBC in the past 3 months. The interviews were 45 minutes long and led by trained moderators via telephone. A final list of 16 potential facilitators and barriers to BRCA testing was selected for the BWS (Table 1 in the electronic supplementary material [ESM]). For the DCE, key attributes associated with PARPi, CDK4/6 inhibitors, immunotherapy, and chemotherapy treatments for eBC were considered. A final list of seven attributes was selected: invasive disease-free survival, targeted treatment, nausea risk, risk of serious side effects (grade 3 and 4; serious side effects [SSE]), regimen, treatment duration, and cost (Table 2 in the ESM). The attributes included two to four levels each, which represented the endpoints of each attribute.

BWS Design

The objective for the BWS was to evaluate the rank order of the relative importance of potential facilitators and barriers to BRCA testing from the patient’s perspective [30]. The 16 attributes were presented in groupings of four over a series of 20 choice tasks, and individuals were asked to select one attribute that would most likely motivate them to get BRCA tested and one attribute that would most likely prevent them from getting BRCA tested (Fig. 1). Each respondent received a unique ordering version of the exercise. A balanced incomplete block design was used to ensure that respondents evaluated a subset of all possible items across multiple-choice sets without overwhelming them with all possible combinations. The design ensures that all items are fairly compared with each other, facilitating more accurate measurement. An a priori sample size calculation was conducted using analytical best–worst for the BWS [31]. Based on this method and assuming α = 0.05 with a margin of error of ± 0.10, a minimum sample size of 154 would produce precise outputs from our BWS design in the aggregate sample. The BWS was programmed using Sawtooth Lighthouse Studio version 9.15.2.

Fig. 1.

Example choice task in the Best–Worst Scaling exercise. Patients were instructed to select one attribute they perceived as most likely to motivate them to get BRCA tested and select one attribute that was most likely to prevent them from getting BRCA testing. Patients completed 20 choice tasks that showed different combinations of attributes. Each attribute was shown approximately five times

DCE Design

A DCE was utilized to evaluate patient preferences for factors related to adjuvant treatment in eBC [32–34]. Over a series of 12 DCE choice tasks, individuals were instructed to select their preferred option from among two hypothetical, unlabeled treatment profiles based on the seven treatment attributes selected or a “no treatment” option (Fig. 2). Each respondent received a randomized, unique ordering version of the exercise. The exercise was created with a balanced overlap design and was programmed in Sawtooth Lighthouse Studio version 9.15.2. The design was generated to optimize overall D-efficiency in terms of (a) level balance (each level is shown approximately an equal number of times); (b) minimal level overlap (levels repeat within the same task); and (c) orthogonality (levels may be evaluated independently of other levels). This helps in distinguishing the preference for different attributes while minimizing confounding effects. The design targeted main effects for the attributes. Given the number of attributes and number of levels within attributes, 864 unique treatment profiles were generated. The final experimental design included 6000 choice tasks split across 500 blocks, with each participant required to complete 12 choice tasks consisting of two unique treatment profiles per task. Twelve choice tasks is consistent with other studies, was considered to not be overly burdensome to individuals, and is robust with respect to the sample size calculation [32]. Simulated data were run to test the precision in which our experimental design could estimate preference weights. The standard errors from the simulated data were <0.04 for every attribute level, which is lower than the recommended guideline that standard errors be <0.05. To determine the minimum sample size for aggregate-level, full-profile DCE modeling, the following formula was used,

$$N\ge 500\frac{L^{\text{max}}}{\mathit{JS}},$$

where L^max reflects the highest number of levels for any one attribute, J is the number of alternatives per choice task, and S represents the number of choice tasks presented to each respondent [35]. Based on this formula, our DCE design (i.e., 12 choice tasks, 2 alternatives per task [excluding the opt out], and a maximum of 4 attribute levels) required a minimum sample size of 83. Given that an opt-out (no treatment) was included, the sample size required increased to around 100, assuming that respondents opt out 20% of the time.

Fig. 2.

Example choice task from the Discrete Choice Experiment to evaluate patients’ preferences for adjuvant treatment versus no treatment

Survey Design

In addition to the BWS and DCE exercises, the broader survey included items to assess individuals’ sociodemographic characteristics, clinical characteristics, and treatment history to provide context on individuals’ backgrounds and experiences. The validity of the survey content was assessed via cognitive debriefing interviews with 10 individuals who met the study criteria [36]. Cognitive interviews were led by a trained member of the study team who used a virtual screen-sharing platform to review survey content and probe individuals on the clarity of survey items. A “think aloud” approach was used, where, as they completed the survey, participants were encouraged to “think aloud,” and asked to describe their understanding of selected survey items, including the attributes/levels in the DCE and BWS. All interviews were recorded, and team members listened in on the interviews. The interviews occurred in three waves between June and July 2023. The first wave included three interviews, whereas the second and third waves included five and two interviews, respectively. Following each wave, the study team met to discuss the participant feedback and whether any revisions should be made to the survey. Ultimately, minor refinements were made to the survey instrument to provide further clarification of the questions. For the BWS, the question text was simplified for ease of interpretation. Specifically, rather than asking respondents to select which statement would most likely and least likely make them get a BRCA test, the question was revised to ask respondents to select which statement would most likely motivate them to get a BRCA test and which statement would most likely prevent them from getting a BRCA test. In addition, one attribute was reworded for ease of interpretation. None of the participants had difficulty interpreting the DCE tasks and wording of the attributes/levels, and no changes were necessary to the attributes/level wording included in the DCE. The experimental designs were not changed for either the BWS or DCE.

The final version of the survey was fielded between October 2023 and March 2024; the average length of the interview was 10.6 minutes. Missing data were prevented by requiring responses for all survey questions, including the BWS and DCE. For potentially sensitive survey items (e.g., educational attainment, race/ethnicity, gender, household income, etc.), individuals were given the option to select “prefer not to say.” Data quality was evaluated through survey programming logic checks at various points in fielding to ensure that responses met survey requirements and that there was no evidence of straight-lining (e.g., selecting the same response on matrix-style items) or speeding (i.e., completing the survey in < 30% of the median length of interview) through the survey. Speeding in the BWS and DCE choice tasks was also specifically evaluated using the same criteria (i.e., completing the activities in < 30% of the median length of time). Finally, for the DCE, root likelihood was also used to evaluate responses for randomness. We simulated 1000 randomly responding participants and used the 80th percentile % root likelihood value as the cutoff for acceptable randomness. This process allowed us to identify respondents who were likely providing random responses and potential bots. In instances where survey responses were flagged for quality concerns, the study team discussed and determined whether the response would be retained in the final dataset or replaced.

Statistical Analyses

Descriptive statistics were used to characterize the overall sample (mean and standard deviation for continuous variables; frequencies and percentages for categorical variables). For the BWS and DCE exercises, Hierarchical Bayes (HB) conditional logit models were estimated [34]. These models can generate utility estimates by combining information at the individual level and data from other respondents to optimize the estimation of coefficients. The model assumes that estimates are normally distributed across respondents in the sample. This model is “hierarchical” because it has two levels. At the higher level, the model assumes that individuals’ preference weights are described by a multivariate normal distribution for the main effects of the attributes, and, at the lower level, the individual’s preference weights are governed by a conditional logistic model. These lower-level individual models can be calibrated using the prior variance and prior degrees of freedom to allow for lower-level models to adhere to the upper-level distribution. These calibration factors were left at their default values (prior variance of 1, 5 degrees of freedom), following guidance from past HB studies [37].

The results of the BWS model consist of relative importance scores that reflect the ratio-scaled utility of each attribute, standardized to a score of 0–100 for reporting. These scores are relative, as participants select the best and the worst options in a comparison of all attributes included in the exercise. To help visualize these estimates as potential barriers or facilitators to BRCA testing, a “neutrality” value was calculated by dividing 100 by the number of attributes included in the BWS (100/16 = 6.3). The relative importance estimates were then centered around the neutrality value, such that a value of 0 represented neutrality (no more likely to motivate or prevent individuals from getting BRCA tested), positive values represented potential facilitators, and negative values represented potential barriers. Estimates are presented with 95% confidence intervals.

The proportion of times an attribute level, or the no treatment option, was selected as part of the preferred treatment profile in the DCE choice task was calculated. Mean preference weights for each attribute level were calculated as estimates of the HB model coefficients using effects coding. Preference weights measure relative preference; only changes between attribute-level estimates and the relative size of those changes across attributes have meaningful interpretations. The absolute difference in the highest and lowest preference estimate for each attribute indicates the magnitude of influence on patient preferences. Ultimately, these preference weights allow for the evaluation of potential trade-offs between utility gains and losses across attributes. Mean preference weights are presented with standard errors and 95% confidence intervals. In addition, the preference weights are used to compute relative importance estimates for each attribute to demonstrate how much difference each attribute could make in the total utility of a treatment. Relative importance estimates were calculated at the patient level by dividing the range of each attribute (utility of most favorable minus least favorable levels) by the sum of ranges of all attributes, and multiplying by 100.

Bivariate comparisons using one-way analysis of variance (analysis of variance; α = 0.05) were performed to examine potential unadjusted differences in patient preferences for BRCA testing and adjuvant treatment relative to age (< 50 years vs. 50–64 years vs. 65 + years), annual household income (≤ $49,999 vs. $50,000 to $99,999 vs. $100,000 +), and family history of cancer (first-degree vs. second-degree or further removed vs. no history).

Data were analyzed using SAS version 9.4 for descriptive statistics and Sawtooth Software Lighthouse Studio version 9.12.0 for the BWS and DCE analyses.

Results

A total of 409 participants met eligibility criteria and completed the survey, of which 50 participants were removed from the final analytic sample after being flagged for data quality issues (n = 37 for providing random responses/potential bots, n = 19 for straight-lining, n = 13 for speeding; note, these sum to > 50 as n = 18 participants were flagged for more than one data quality issue). Therefore, the final analytic sample included 359 participants, which represented 8.3% of the 4,321 panelists who accessed the survey screener. Participants had a median age of 63.0 years. The largest proportion of individuals were retired (n = 169; 47.1%), married (n = 180; 50.1%), lived in a major metropolitan area (n = 155; 43.2%), and had a net income between $50,000 and $99,000 annually (n = 141; 39.3%). Among individuals, over half were HR-positive (n = 202; 56.3%), and approximately half were not tested for BRCAm (n = 220; 61.3%). Almost all individuals previously received surgery for their breast cancer (n = 340; 94.7%), over one-third of individuals (n = 137; 38.2%) did not receive surgery, and 61.5% received chemotherapy. Additional sociodemographic, clinical, and treatment characteristics are shown in Table 1.

Table 1.

Sociodemographic, clinical, and treatment characteristics

	Sample N = 359
Age, years	63 (25–91)
Race; n (%)
American Indian or Alaskan Native	31 (8.6)
Asian	20 (5.6)
Black or African American	29 (8.1)
White	277 (77.2)
Other/prefer not to say	9 (2.5)
Ethnicity
Cuban	11 (3.1)
Jewish	18 (5.0)
Mexican, Mexican American, Chicano	15 (4.2)
Puerto Rican	5 (1.4)
Other Hispanic, Latino, or Spanish origin	39 (10.9)
Other ethnicity or heritage not listed	73 (20.3)
None of the above/prefer not to say	198 (55.1)
University Degree or Higher [yes]	153 (42.6)
Current Work Situation
Currently employed	110 (30.6)
Retired	169 (47.1)
Othera	62 (17.3)
Prefer not to answer	18 (5.0)
Marital Status
Single (never married)	28 (7.8)
Married/living with partner (never married)	180 (50.1)
Separated/divorced/widowed	136 (37.9)
Prefer not to answer	15 (4.2)
Location of residence
Major metropolitan/urban area	155 (43.2)
Suburb of a large city	89 (24.8)
Small city/rural or small town	103 (28.6)
Prefer not to answer	12 (3.3)
Household income
< $50,000	100 (27.9)
$50,000–99,999	141 (39.3)
$100,000–149,999	76 (21.2)
≥ $150,000	36 (10.0)
Prefer not to answer	6 (1.7)
Health Insurance (multi-select)
Private/commercial	126 (35.1)
Medicare	259 (72.2)
Veteran’s Administration/CHAMPUS/TRICARE	11 (3.1)
Other not listed	6 (1.7)
Prefer not to say	52 (14.5)
Year diagnosed	2020 (1985–2023)
Stage at Diagnosis
Stage 0	51 (14.2)
Stage 1	150 (41.8)
Stage 2	112 (31.2)
Stage 3	42 (11.7)
I’m not sure	4 (1.1)
gBRCA testing status
Untestedb	220 (61.3)
gBRCA mutated	139 (38.7)
HR status
HR positive	202 (56.3)
HR negative	133 (37.0)
Not sure	24 (6.7)
Received chemotherapy [Yes]c	221 (61.5)
Received surgery to treat breast cancer
Yes	340 (94.7)
No, but schedule or plan to have surgery	19 (5.3)

Data are presented as median (minimum, maximum) or n (%) unless otherwise indicated

gBRCA, germline BReast CAncer gene; HR, hormone receptor

^aIncludes the following response options: not employed, disability, temporary leave of absence, stay-at-home, and temporarily laid-off

^bIncludes 192 patients who said “no, I have never tested for a BRCA mutation,” three who said “Yes, but I am unsure of the results,” and 25 who said “I’m not sure if I was tested”

^cIncludes neoadjuvant, adjuvant, or both

Facilitators and Barriers to BRCA Testing

Figure 3 illustrates the BWS scores for facilitators and barriers to getting BRCA testing centered around the neutrality value of 6.3. The scores for the 16 factors sum to 0; factors with positive values were more likely to drive individuals to get a BRCA test, whereas those with negative scores were more likely to prevent individuals from getting a BRCA test. Relative to the other factors in the BWS, the top three facilitators for BRCA testing were receiving a physician’s recommendation, determining eligibility for targeted therapy that may prevent or delay metastasis, and having no out-of-pocket cost (OOPC) for testing. OOPC of $250, concerns that testing may cause the respondent anxiety, and concerns of testing potentially delaying treatment by 3–4 weeks were the top three barriers.

Fig. 3.

Estimated mean BWS scores (relative importance) of potential barriers and facilitators to BRCA testing. BWS, best-worst-scaling exercise. Relative importance estimates were estimated at the individual level from the BWS then summarized as mean scores in the aggregate sample. Here, scores are centered around the “neutrality” value (6.3), which was derived by dividing 100 by the number of attributes included in the BWS (16). Positive values indicate potential facilitators to BRCA testing, whereas negative values indicate potential barriers. Estimates are shown with 95% confidence intervals and sum to 0

In the subgroup analysis performed to evaluate unadjusted differences in perceived facilitators/barriers to BRCA testing, few differences were identified relative to age (Table 3 in the ESM) and a family history of cancer (Table 4 in the ESM), whereas greater variation was observed relative to household income (Table 5 in the ESM). In general, testing to determine eligibility for preventive treatments, physician recommendation, and having no cost for BRCA testing were among the top facilitators to BRCA testing, regardless of subgroups, with slight variations in the rank ordering and magnitude of importance. Individuals with the lowest household income found having as much information about their breast cancer as possible to be a more important facilitator than there being no cost to BRCA testing and individuals with moderate household income levels found having the ability to share BRCA testing results with their family members who may be impacted equally as important as determining eligibility for preventive treatments (Table 5 in the ESM). The top three barriers were also consistent across all subgroups and included the $250 out-of-pocket cost for BRCA testing, concerns that testing may cause the respondent anxiety, and potential delays in treatment due to testing, with slight variation in the rank ordering and magnitude of importance across groups. Individuals with the lowest household income also noted that BRCA testing results potentially leading to denial of life insurance or higher premiums was a perceived barrier to BRCA testing that was equally as important as testing, causing them anxiety and potentially delaying treatment (Table 5 in the ESM).

Preferences for Adjuvant Treatment

Results from the DCE exercise showed that when presented with hypothetical adjuvant treatment profiles, participants selected one of the adjuvant treatment profiles over the no treatment option 77.6% of the time, regardless of the combination of treatment attributes shown. Reducing OOPC from $900 per month to $0 per month (|− 5.00 to 3.48| = 8.48) had the greatest influence on preferences for adjuvant treatment relative to the other attributes assessed (Fig. 4, Table 2). Additionally, reducing the risk of serious side effects from 77 to 24% (|− 0.89 to 1.02| = 1.91) and being able to receive a targeted treatment based on the presence of a BRCA mutation or not (|0.64 to − 0.64| = 1.28) also influenced preference to a greater degree than other attributes assessed (Fig. 4, Table 2).

Fig. 4.

Estimated mean preference weights from DCE evaluating patient preferences for adjuvant therapy. DCE, discrete choice experiment; IDFS, invasive disease-free survival; SE, side effect. In each choice task, patients were shown, “Please imagine that you have just had your surgery for your breast cancer, and you have the following treatment options to choose from after your surgery. Which option would you most prefer?” Preference weights were estimated at the individual-level and summarized as means in the aggregate sample. Preference weights should not be interpreted by themselves. Instead, the magnitude of change within one attribute should be compared to change within another attribute; the greater the magnitude of change reflects greater influence—either positive or negative—on treatment selection. All preference weights of levels within an attribute sum to 0. p Values for all preference weights were < 0.05

Table 2.

Estimated mean attribute-level preference weights from the discrete-choice experiment (DCE) evaluating patient preferences for adjuvant therapy

Preferences	b	SE	95% CI
Out-of-pocket cost
$0 per month	3.48	0.15	3.18–3.79
$45 per month	2.53	0.08	2.37–2.69
$377 per month	− 1.01	0.06	− 1.14 to − 0.88
$900 per month	− 5.00	0.17	− 5.34 to − 4.67
Serious side effect (grade 3/4)
24 out of 100 patients have a serious side effect that may require medical intervention or hospitalization	1.02	0.03	0.95–1.08
50 out of 100 patients have a serious side effect that may require medical intervention or hospitalization	− 0.13	0.02	− 0.16 to − 0.09
77 out of 100 patients have a serious side effect that may require medical intervention or hospitalization	− 0.89	0.03	− 0.96 to − 0.83
Targeted treatment
Treatment is targeted based on a BRCA mutation	0.64	0.04	0.57–0.71
Treatment is not targeted based on a BRCA mutation	− 0.64	0.04	− 0.71 to − 0.57
Nausea risk
27 out of 100 patients will have mild to moderate nausea	0.37	0.02	0.32–0.42
60 out of 100 patients will have mild to moderate nausea	0.04	0.02	0.01–0.08
98 out of 100 patients will have mild to moderate nausea	− 0.41	0.02	− 0.45 to − 0.37
Invasive disease-free survival
82 out of 100 patients are disease free at 3-year follow-up	− 0.20	0.02	− 0.23 to − 0.16
89 out of 100 patients are disease free at 3-year follow-up	0.20	0.02	0.16–0.23
Regimen
Oral pills twice per day	− 0.03	0.02	− 0.06 to 0
Oral pills twice per day; one week break every two weeks	0.16	0.02	0.12–0.2
30-min IV infusion every 3 or 6 weeks	− 0.13	0.02	− 0.16 to − 0.09
Duration
Treatment is taken for one year	0.10	0.01	0.07–0.12
Treatment is taken for two years	− 0.10	0.01	− 0.12 to − 0.07
No treatment
77 out of 100 patients are disease free at 3-year follow-up	− 2.34	0.29	− 2.9 to − 1.78

In each choice task, patients were shown, “Please imagine that you have just had your surgery for your breast cancer, and you have the following treatment options to choose from after your surgery. Which option would you most prefer?” Preference weights were estimated at the individual-level and summarized as means in the aggregate sample. Preference weights should not be interpreted by themselves. Instead, the magnitude of change within one attribute should be compared to change within another attribute; the greater the magnitude of change reflects greater influence—either positive or negative—on treatment selection. All preference weights of levels within an attribute sum to 0

b, mean preference weight; CI, confidence interval; SE, standard error

No differences were found in patient preferences for adjuvant treatment attributes relative to individuals’ family history of cancer. However, the relative importance of treatment regimen preferences varied by age, such that older individuals found regimen significantly more important than younger individuals (p < 0.05; Fig. 1A in the ESM). Moreover, the relative importance of reducing the risk of SSEs was significantly higher among those who made the least income compared to those who made between $50,000 and $99,999 in the past year (p < 0.05; Fig. 1B in the ESM).

Discussion

This study explored the factors influencing patient preferences for BRCA testing and adjuvant therapy in individuals with HER2-negative eBC. Understanding these preferences is important, as access to olaparib for high-risk, gBRCAm HER2-negative eBC is often limited by the underutilization of BRCA testing [17, 18, 38]. Our study demonstrated that individuals are willing to consider adjuvant therapy and that a key motivator for undergoing BRCA testing is to better inform their treatment decisions because testing results may show that an individual is eligible for a targeted treatment that may prevent or delay cancer metastases. This openness underscores the importance of determining a patient’s BRCA status to ensure individuals have access to the most effective therapy.

Results from our BWS exercise showed that several factors explained individuals’ preferences for BRCA testing. Facilitators included learning about preventive therapy options, receiving a physician recommendation for testing, having as much information about breast cancer as possible, and learning about potential BRCA-targeted treatments. These findings suggest that when individuals are provided with clear, comprehensive information about the potential benefits of BRCA testing and its role in informing treatment decisions, they may be more likely to see genetic testing as an essential component of their care. Previous research indicates that physicians play a key role in encouraging patients to pursue BRCA testing. A survey-based study found that the uptake of BRCA testing is predominantly driven by physician referrals, with few patients seeking testing without a physician’s suggestion [39]. These results highlight the need for clinicians to engage patients in comprehensive discussions about the advantages of BRCA testing, ensuring that patients understand how testing can impact a patient’s treatment plan [40]. By fostering these conversations, clinicians can empower patients to make well-informed healthcare decisions and help them feel more involved in their care [41].

Conversely, multiple reasons were given for not pursuing BRCA testing, with OOPC being the largest barrier. This finding is consistent with a survey-based study that examined the challenges of gBRCA testing from the perspectives of individuals with HR+/HER2-negative eBC, healthcare professionals, and payers [17]. The study showed that high OOPC for gBRCA testing and genetic counseling, as well as coverage for genetic counseling, were the top challenges for individuals [17]. These results should interest payers, who may be unaware of the extent to which individuals consider cost as a barrier. Indeed, within the current system where payment is required, individuals may be choosing to forego BRCA testing because of these requirements. Individuals who refuse BRCA testing due to cost may miss the opportunity to receive the best treatment possible, which may potentially lower long-term costs for payers and contribute to a better prognosis [11, 42]. Therefore, payers may want to consider the potential gains associated with patients receiving the best care possible versus requiring beneficiaries to pay for genetic testing.

Other barriers to BRCA testing included potential delays in treatment, concerns about denial of life insurance or higher life insurance premiums, and the potential for BRCA test results to cause anxiety. Fears of genetic discrimination among life insurers are well-documented and contribute to many individuals opting out of genetic testing [43–48]. While the Genetic Information Nondiscrimination Act of 2008 protects against genetic discrimination in health insurance, it does not extend to life insurance, which allows insurers to use genetic results in underwriting [49]. Currently, Florida is the only state that prohibits such use [50]. As a result, individuals may be hesitant to pursue BRCA testing due to concerns that a positive result could lead to higher life insurance premiums or even denial of coverage, despite its clinical benefits [51–53].

Our DCE results showed that over three-quarters of individuals prefer adjuvant therapy over no treatment, regardless of the combination of treatment attributes shown. This finding aligns with previous studies demonstrating that most individuals with eBC consider adjuvant therapy worthwhile, despite the associated side effects and inconvenience [54–56]. In our study, the most influential factor shaping individuals’ preferences for adjuvant therapy was a reduction in OOPC. The financial burden faced by individuals with breast cancer due to the long-term need for multidisciplinary care and high treatment costs is widely recognized [57, 58]. Therefore, physicians should carefully consider individuals' financial concerns when discussing treatment options.

Cost was not the only factor that influenced patient preferences for adjuvant treatment. Individuals also showed a preference for treatments that lowered the risk of SSEs, used a targeted approach, and lowered the risk of nausea. These findings differ from prior research examining the factors influencing treatment decisions among individuals with breast cancer [59, 60], which found that while cost is an important factor, treatment effectiveness and quality of life were prioritized. This discrepancy may reflect differences in study populations, as the prior studies included patients of all stages of breast cancer, whereas our study focused specifically on patients with HER2-negative eBC. The discrepancy may also be due to differences in study designs, including the attributes and levels included in each study, as preference weights only have meaning within the context of the other attributes and levels. It also may reflect the nuanced and complex nature of patient decision-making in breast cancer treatment. These results suggest that patients are open to receiving adjuvant treatments that target their specific cancer, though patient–physician conversations regarding potential targeted therapies should be tailored to individual patient preferences and include transparency regarding the safety profiles and potential financial costs of treatments.

Our study observed slight differences in patient preferences within certain subgroups. While results from the BWS showed a similar rank-ordering for facilitators and barriers to BRCA testing across subgroups (age, household income, and family history of cancer), some individuals placed greater emphasis on certain barriers and facilitators than others. Specifically, individuals under 50 years prioritized learning about eligibility for preventive therapy, those aged 50–64 years valued the absence of OOPC for BRCA testing, and individuals 65 years and older considered a physician’s recommendation to be the most significant factor. Furthermore, DCE results found that reducing OOPC was the top priority across all age and income groups when selecting an adjuvant therapy. However, older individuals prioritized treatment administration, while lower-income individuals valued reducing SSEs more than other subgroups. Our results echo previous findings suggesting that age is an influential factor in patient preferences for cancer care, with evidence to suggest that older patients prefer less aggressive treatment regimens, limited hospital stays [61, 62], and may rely more on physician recommendations for treatment decisions [63]. Moreover, income has long been documented as a driver of disparities in healthcare access, which can influence both patient and physician treatment decisions [64]. Our findings build upon this evidence by illustrating the diverse experiences, knowledge levels, and attitudes of individuals with eBC, underscoring the importance of shared decision-making to address the diverse needs of this patient population. When shared decision-making is prioritized, individuals report greater satisfaction with their care, improved knowledge about their treatment options, reduced decision regret, stronger commitment to chosen treatments, and improved perceptions of coordinated care [65–68]. Together, these findings highlight the need for personalized care and decision-making processes in eBC, ensuring that treatment choices reflect the unique priorities of each patient.

Limitations

The use of an online patient-reported survey with a convenience sample may introduce certain limitations. This approach may result in participants who are more actively involved in managing their disease, as the participants were selected from a panel’s list of individuals who have expressed a willingness to participate in health-related surveys. Additionally, it may underrepresent individuals who lack internet access, individuals who are uncomfortable using online tools, and individuals with severe comorbidities or disabilities who may be less likely to participate. The survey did not include men, and thus their perspectives are not represented. The self-reported nature of the survey may have also resulted in potential bias, such as inaccurate recall and false reporting. Further, eligibility criteria were self-reported and not verified by an independent source, such as electronic medical records.

Due to the timing of this research, this study was limited in its ability to distinguish between individuals’ perspectives on targeted treatment options for gBRCA breast cancer and the availability of such treatments. Olaparib was approved by the US Food and Drug Administration in March 2022; however, our study included individuals who were diagnosed and received treatment for breast cancer prior to the approval of olaparib [9]. Restricting the sample to only individuals diagnosed following the approval of olaparib was not feasible for this study; therefore, we are unable to determine whether the availability of targeted therapies influenced individuals’ perceptions of BRCA testing and adjuvant treatment.

The BWS and DCE methodology used in our study involved respondents choosing between hypothetical treatment profiles and testing characteristics. As such, results from this study may not reflect real-world testing and treatment decisions, which could be influenced by other factors not captured in the survey. It should also be noted that this survey is based on a US setting, and preferences may differ in other healthcare systems and regions. Lastly, the sequence of choice tasks for BRCA testing (BWS) to preferences for adjuvant therapy (DCE) may introduce the potential for ordering effects and choice-task fatigue to occur. While previous research has identified learning and fatigue effects in choice experiments with numerous choice sets or multiple-choice tasks, these effects were typically modest [69, 70]. Nonetheless, this potential limitation warrants consideration when interpreting the results.

Conclusion

Individuals indicated a key motivator for undergoing BRCA testing was to inform treatment decisions, especially in situations where testing results could inform targeted treatment. However, OOPC likely acts as a barrier to patient choice for testing. Decisions to receive adjuvant therapy were most influenced by OOPC, followed by the tolerability of the treatment and the ability to receive a targeted therapy. Noting that individual preferences for both BRCA testing and adjuvant therapy likely vary relative to patient factors, such as age, income, and family history of cancer, our findings underscore the value of comprehensive patient–provider discussions that foster shared decision-making. In particular, discussions regarding the benefits of testing, including potential eligibility for targeted therapies with a favorable safety profile, may be an effective strategy for increasing BRCA testing while encouraging patients to share their goals for eBC treatment. Similarly, selection of an effective adjuvant treatment likely requires balanced conversations that consider medication tolerability and financial burden to the patient.

Supplementary Information

Below is the link to the electronic supplementary material.

Funding

This study was funded by Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA, and AstraZeneca UK Ltd, who are co-developing Olaparib.

Declarations

Conflicts of interest

Kathryn Mishkin: Employee of Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA (MSD). Qixin Li: Employee of AstraZeneca. Jagadeswara Rao Earla: Employee of MSD. Jaime Mejia: Employee of MSD. Kim M. Hirshfield: Employee of MSD. Kathryn Krupsky: Employee of Oracle Life Sciences, which received funding from MSD to conduct this research. Josh Lankin: Employee of Oracle Life Sciences, which received funding from MSD to conduct this research. Kathleen Beusterien: Employee of Oracle Life Sciences, which received funding from MSD to conduct this research. Emily Mulvihill: Employee of Oracle Life Sciences, which received funding from MSD to conduct this research. Ryan Honomichl: Employee of Oracle Life Sciences, which received funding from MSD to conduct this research. Alexandra Gordon: Employee of Oracle Life Sciences, which received funding from MSD to conduct this research. Xiaoqing Xu: Employee of AstraZeneca.

Availability of data and material

The data supporting the findings of this study are available upon request.

Ethics approval

The authors state that this study received exemption status from full or expedited ethical review by the Sterling Institutional Review Board (IRB ID: 10632-EMulvihill) on December 22, 2022, and followed the principles outlined in the Declaration of Helsinki for all human or experimental investigations.

Consent to participate

Individuals indicated their interest in participating and gave informed consent electronically.

Consent to publish

Not applicable

Code availability

All software applications used are commercially available.

Author contributions

Kathryn Mishkin: conceptualization; writing—review and editing; supervision. Qixin Li: conceptualization; writing—review and editing. Jagadeswara Rao Earla: conceptualization; writing—review and editing; supervision. Jaime Mejia: writing—review and editing. Kim M. Hirshfield: writing—review and editing. Kathryn Krupsky: methodology; validation; formal analysis; writing—original draft; project administration. Josh Lankin: conceptualization; methodology; writing—review and editing; project administration. Kathleen Beusterien: conceptualization; methodology; validation; writing—review and editing. Emily Mulvihill: conceptualization; methodology; writing—review and editing; project administration. Ryan Honomichl: methodology; validation; formal analysis; data curation; writing—review and editing. Alexandra Gordon: methodology; validation; formal analysis; data curation; writing—review and editing. Xiaoqing Xu: conceptualization; writing—review and editing. All authors read and approved the final version of the manuscript.